diff --git a/vendor/github.com/google/go-cmp/cmp/compare.go b/vendor/github.com/google/go-cmp/cmp/compare.go
index c9a63ce..86d0903 100644
--- a/vendor/github.com/google/go-cmp/cmp/compare.go
+++ b/vendor/github.com/google/go-cmp/cmp/compare.go
@@ -1,11 +1,15 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // Package cmp determines equality of values.
 //
 // This package is intended to be a more powerful and safer alternative to
 // reflect.DeepEqual for comparing whether two values are semantically equal.
+// It is intended to only be used in tests, as performance is not a goal and
+// it may panic if it cannot compare the values. Its propensity towards
+// panicking means that its unsuitable for production environments where a
+// spurious panic may be fatal.
 //
 // The primary features of cmp are:
 //
@@ -86,6 +90,52 @@
 // If there is a cycle, then the pointed at values are considered equal
 // only if both addresses were previously visited in the same path step.
 func Equal(x, y interface{}, opts ...Option) bool {
+	s := newState(opts)
+	s.compareAny(rootStep(x, y))
+	return s.result.Equal()
+}
+
+// Diff returns a human-readable report of the differences between two values:
+// y - x. It returns an empty string if and only if Equal returns true for the
+// same input values and options.
+//
+// The output is displayed as a literal in pseudo-Go syntax.
+// At the start of each line, a "-" prefix indicates an element removed from x,
+// a "+" prefix to indicates an element added from y, and the lack of a prefix
+// indicates an element common to both x and y. If possible, the output
+// uses fmt.Stringer.String or error.Error methods to produce more humanly
+// readable outputs. In such cases, the string is prefixed with either an
+// 's' or 'e' character, respectively, to indicate that the method was called.
+//
+// Do not depend on this output being stable. If you need the ability to
+// programmatically interpret the difference, consider using a custom Reporter.
+func Diff(x, y interface{}, opts ...Option) string {
+	s := newState(opts)
+
+	// Optimization: If there are no other reporters, we can optimize for the
+	// common case where the result is equal (and thus no reported difference).
+	// This avoids the expensive construction of a difference tree.
+	if len(s.reporters) == 0 {
+		s.compareAny(rootStep(x, y))
+		if s.result.Equal() {
+			return ""
+		}
+		s.result = diff.Result{} // Reset results
+	}
+
+	r := new(defaultReporter)
+	s.reporters = append(s.reporters, reporter{r})
+	s.compareAny(rootStep(x, y))
+	d := r.String()
+	if (d == "") != s.result.Equal() {
+		panic("inconsistent difference and equality results")
+	}
+	return d
+}
+
+// rootStep constructs the first path step. If x and y have differing types,
+// then they are stored within an empty interface type.
+func rootStep(x, y interface{}) PathStep {
 	vx := reflect.ValueOf(x)
 	vy := reflect.ValueOf(y)
 
@@ -108,33 +158,7 @@
 		t = vx.Type()
 	}
 
-	s := newState(opts)
-	s.compareAny(&pathStep{t, vx, vy})
-	return s.result.Equal()
-}
-
-// Diff returns a human-readable report of the differences between two values.
-// It returns an empty string if and only if Equal returns true for the same
-// input values and options.
-//
-// The output is displayed as a literal in pseudo-Go syntax.
-// At the start of each line, a "-" prefix indicates an element removed from x,
-// a "+" prefix to indicates an element added to y, and the lack of a prefix
-// indicates an element common to both x and y. If possible, the output
-// uses fmt.Stringer.String or error.Error methods to produce more humanly
-// readable outputs. In such cases, the string is prefixed with either an
-// 's' or 'e' character, respectively, to indicate that the method was called.
-//
-// Do not depend on this output being stable. If you need the ability to
-// programmatically interpret the difference, consider using a custom Reporter.
-func Diff(x, y interface{}, opts ...Option) string {
-	r := new(defaultReporter)
-	eq := Equal(x, y, Options(opts), Reporter(r))
-	d := r.String()
-	if (d == "") != eq {
-		panic("inconsistent difference and equality results")
-	}
-	return d
+	return &pathStep{t, vx, vy}
 }
 
 type state struct {
@@ -352,7 +376,7 @@
 // assuming that T is assignable to R.
 // Otherwise, it returns the input value as is.
 func sanitizeValue(v reflect.Value, t reflect.Type) reflect.Value {
-	// TODO(dsnet): Workaround for reflect bug (https://golang.org/issue/22143).
+	// TODO(≥go1.10): Workaround for reflect bug (https://golang.org/issue/22143).
 	if !flags.AtLeastGo110 {
 		if v.Kind() == reflect.Interface && v.IsNil() && v.Type() != t {
 			return reflect.New(t).Elem()
@@ -362,6 +386,7 @@
 }
 
 func (s *state) compareStruct(t reflect.Type, vx, vy reflect.Value) {
+	var addr bool
 	var vax, vay reflect.Value // Addressable versions of vx and vy
 
 	var mayForce, mayForceInit bool
@@ -383,6 +408,7 @@
 				// For retrieveUnexportedField to work, the parent struct must
 				// be addressable. Create a new copy of the values if
 				// necessary to make them addressable.
+				addr = vx.CanAddr() || vy.CanAddr()
 				vax = makeAddressable(vx)
 				vay = makeAddressable(vy)
 			}
@@ -393,6 +419,7 @@
 				mayForceInit = true
 			}
 			step.mayForce = mayForce
+			step.paddr = addr
 			step.pvx = vax
 			step.pvy = vay
 			step.field = t.Field(i)
diff --git a/vendor/github.com/google/go-cmp/cmp/export_panic.go b/vendor/github.com/google/go-cmp/cmp/export_panic.go
index dd03235..5ff0b42 100644
--- a/vendor/github.com/google/go-cmp/cmp/export_panic.go
+++ b/vendor/github.com/google/go-cmp/cmp/export_panic.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // +build purego
 
@@ -10,6 +10,6 @@
 
 const supportExporters = false
 
-func retrieveUnexportedField(reflect.Value, reflect.StructField) reflect.Value {
+func retrieveUnexportedField(reflect.Value, reflect.StructField, bool) reflect.Value {
 	panic("no support for forcibly accessing unexported fields")
 }
diff --git a/vendor/github.com/google/go-cmp/cmp/export_unsafe.go b/vendor/github.com/google/go-cmp/cmp/export_unsafe.go
index 57020e2..21eb548 100644
--- a/vendor/github.com/google/go-cmp/cmp/export_unsafe.go
+++ b/vendor/github.com/google/go-cmp/cmp/export_unsafe.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // +build !purego
 
@@ -17,9 +17,19 @@
 // a struct such that the value has read-write permissions.
 //
 // The parent struct, v, must be addressable, while f must be a StructField
-// describing the field to retrieve.
-func retrieveUnexportedField(v reflect.Value, f reflect.StructField) reflect.Value {
-	// See https://github.com/google/go-cmp/issues/167 for discussion of the
-	// following expression.
-	return reflect.NewAt(f.Type, unsafe.Pointer(uintptr(unsafe.Pointer(v.UnsafeAddr()))+f.Offset)).Elem()
+// describing the field to retrieve. If addr is false,
+// then the returned value will be shallowed copied to be non-addressable.
+func retrieveUnexportedField(v reflect.Value, f reflect.StructField, addr bool) reflect.Value {
+	ve := reflect.NewAt(f.Type, unsafe.Pointer(uintptr(unsafe.Pointer(v.UnsafeAddr()))+f.Offset)).Elem()
+	if !addr {
+		// A field is addressable if and only if the struct is addressable.
+		// If the original parent value was not addressable, shallow copy the
+		// value to make it non-addressable to avoid leaking an implementation
+		// detail of how forcibly exporting a field works.
+		if ve.Kind() == reflect.Interface && ve.IsNil() {
+			return reflect.Zero(f.Type)
+		}
+		return reflect.ValueOf(ve.Interface()).Convert(f.Type)
+	}
+	return ve
 }
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/diff/debug_disable.go b/vendor/github.com/google/go-cmp/cmp/internal/diff/debug_disable.go
index fe98dcc..1daaaac 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/diff/debug_disable.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/diff/debug_disable.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // +build !cmp_debug
 
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/diff/debug_enable.go b/vendor/github.com/google/go-cmp/cmp/internal/diff/debug_enable.go
index 597b6ae..4b91dbc 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/diff/debug_enable.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/diff/debug_enable.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // +build cmp_debug
 
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/diff/diff.go b/vendor/github.com/google/go-cmp/cmp/internal/diff/diff.go
index 3d2e426..bc196b1 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/diff/diff.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/diff/diff.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // Package diff implements an algorithm for producing edit-scripts.
 // The edit-script is a sequence of operations needed to transform one list
@@ -12,6 +12,13 @@
 // is more important than obtaining a minimal Levenshtein distance.
 package diff
 
+import (
+	"math/rand"
+	"time"
+
+	"github.com/google/go-cmp/cmp/internal/flags"
+)
+
 // EditType represents a single operation within an edit-script.
 type EditType uint8
 
@@ -112,6 +119,8 @@
 	return r.NumSame+1 >= r.NumDiff
 }
 
+var randBool = rand.New(rand.NewSource(time.Now().Unix())).Intn(2) == 0
+
 // Difference reports whether two lists of lengths nx and ny are equal
 // given the definition of equality provided as f.
 //
@@ -177,6 +186,11 @@
 	// approximately the square-root of the search budget.
 	searchBudget := 4 * (nx + ny) // O(n)
 
+	// Running the tests with the "cmp_debug" build tag prints a visualization
+	// of the algorithm running in real-time. This is educational for
+	// understanding how the algorithm works. See debug_enable.go.
+	f = debug.Begin(nx, ny, f, &fwdPath.es, &revPath.es)
+
 	// The algorithm below is a greedy, meet-in-the-middle algorithm for
 	// computing sub-optimal edit-scripts between two lists.
 	//
@@ -194,20 +208,26 @@
 	//	frontier towards the opposite corner.
 	//	• This algorithm terminates when either the X coordinates or the
 	//	Y coordinates of the forward and reverse frontier points ever intersect.
-	//
+
 	// This algorithm is correct even if searching only in the forward direction
 	// or in the reverse direction. We do both because it is commonly observed
 	// that two lists commonly differ because elements were added to the front
 	// or end of the other list.
 	//
-	// Running the tests with the "cmp_debug" build tag prints a visualization
-	// of the algorithm running in real-time. This is educational for
-	// understanding how the algorithm works. See debug_enable.go.
-	f = debug.Begin(nx, ny, f, &fwdPath.es, &revPath.es)
-	for {
+	// Non-deterministically start with either the forward or reverse direction
+	// to introduce some deliberate instability so that we have the flexibility
+	// to change this algorithm in the future.
+	if flags.Deterministic || randBool {
+		goto forwardSearch
+	} else {
+		goto reverseSearch
+	}
+
+forwardSearch:
+	{
 		// Forward search from the beginning.
 		if fwdFrontier.X >= revFrontier.X || fwdFrontier.Y >= revFrontier.Y || searchBudget == 0 {
-			break
+			goto finishSearch
 		}
 		for stop1, stop2, i := false, false, 0; !(stop1 && stop2) && searchBudget > 0; i++ {
 			// Search in a diagonal pattern for a match.
@@ -242,10 +262,14 @@
 		} else {
 			fwdFrontier.Y++
 		}
+		goto reverseSearch
+	}
 
+reverseSearch:
+	{
 		// Reverse search from the end.
 		if fwdFrontier.X >= revFrontier.X || fwdFrontier.Y >= revFrontier.Y || searchBudget == 0 {
-			break
+			goto finishSearch
 		}
 		for stop1, stop2, i := false, false, 0; !(stop1 && stop2) && searchBudget > 0; i++ {
 			// Search in a diagonal pattern for a match.
@@ -280,8 +304,10 @@
 		} else {
 			revFrontier.Y--
 		}
+		goto forwardSearch
 	}
 
+finishSearch:
 	// Join the forward and reverse paths and then append the reverse path.
 	fwdPath.connect(revPath.point, f)
 	for i := len(revPath.es) - 1; i >= 0; i-- {
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/flags/flags.go b/vendor/github.com/google/go-cmp/cmp/internal/flags/flags.go
index a9e7fc0..d8e459c 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/flags/flags.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/flags/flags.go
@@ -1,6 +1,6 @@
 // Copyright 2019, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package flags
 
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/flags/toolchain_legacy.go b/vendor/github.com/google/go-cmp/cmp/internal/flags/toolchain_legacy.go
index 01aed0a..82d1d7f 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/flags/toolchain_legacy.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/flags/toolchain_legacy.go
@@ -1,6 +1,6 @@
 // Copyright 2019, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // +build !go1.10
 
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/flags/toolchain_recent.go b/vendor/github.com/google/go-cmp/cmp/internal/flags/toolchain_recent.go
index c0b667f..8646f05 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/flags/toolchain_recent.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/flags/toolchain_recent.go
@@ -1,6 +1,6 @@
 // Copyright 2019, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // +build go1.10
 
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/function/func.go b/vendor/github.com/google/go-cmp/cmp/internal/function/func.go
index ace1dbe..d127d43 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/function/func.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/function/func.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // Package function provides functionality for identifying function types.
 package function
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/value/name.go b/vendor/github.com/google/go-cmp/cmp/internal/value/name.go
new file mode 100644
index 0000000..b6c12ce
--- /dev/null
+++ b/vendor/github.com/google/go-cmp/cmp/internal/value/name.go
@@ -0,0 +1,157 @@
+// Copyright 2020, The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package value
+
+import (
+	"reflect"
+	"strconv"
+)
+
+// TypeString is nearly identical to reflect.Type.String,
+// but has an additional option to specify that full type names be used.
+func TypeString(t reflect.Type, qualified bool) string {
+	return string(appendTypeName(nil, t, qualified, false))
+}
+
+func appendTypeName(b []byte, t reflect.Type, qualified, elideFunc bool) []byte {
+	// BUG: Go reflection provides no way to disambiguate two named types
+	// of the same name and within the same package,
+	// but declared within the namespace of different functions.
+
+	// Named type.
+	if t.Name() != "" {
+		if qualified && t.PkgPath() != "" {
+			b = append(b, '"')
+			b = append(b, t.PkgPath()...)
+			b = append(b, '"')
+			b = append(b, '.')
+			b = append(b, t.Name()...)
+		} else {
+			b = append(b, t.String()...)
+		}
+		return b
+	}
+
+	// Unnamed type.
+	switch k := t.Kind(); k {
+	case reflect.Bool, reflect.String, reflect.UnsafePointer,
+		reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64,
+		reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr,
+		reflect.Float32, reflect.Float64, reflect.Complex64, reflect.Complex128:
+		b = append(b, k.String()...)
+	case reflect.Chan:
+		if t.ChanDir() == reflect.RecvDir {
+			b = append(b, "<-"...)
+		}
+		b = append(b, "chan"...)
+		if t.ChanDir() == reflect.SendDir {
+			b = append(b, "<-"...)
+		}
+		b = append(b, ' ')
+		b = appendTypeName(b, t.Elem(), qualified, false)
+	case reflect.Func:
+		if !elideFunc {
+			b = append(b, "func"...)
+		}
+		b = append(b, '(')
+		for i := 0; i < t.NumIn(); i++ {
+			if i > 0 {
+				b = append(b, ", "...)
+			}
+			if i == t.NumIn()-1 && t.IsVariadic() {
+				b = append(b, "..."...)
+				b = appendTypeName(b, t.In(i).Elem(), qualified, false)
+			} else {
+				b = appendTypeName(b, t.In(i), qualified, false)
+			}
+		}
+		b = append(b, ')')
+		switch t.NumOut() {
+		case 0:
+			// Do nothing
+		case 1:
+			b = append(b, ' ')
+			b = appendTypeName(b, t.Out(0), qualified, false)
+		default:
+			b = append(b, " ("...)
+			for i := 0; i < t.NumOut(); i++ {
+				if i > 0 {
+					b = append(b, ", "...)
+				}
+				b = appendTypeName(b, t.Out(i), qualified, false)
+			}
+			b = append(b, ')')
+		}
+	case reflect.Struct:
+		b = append(b, "struct{ "...)
+		for i := 0; i < t.NumField(); i++ {
+			if i > 0 {
+				b = append(b, "; "...)
+			}
+			sf := t.Field(i)
+			if !sf.Anonymous {
+				if qualified && sf.PkgPath != "" {
+					b = append(b, '"')
+					b = append(b, sf.PkgPath...)
+					b = append(b, '"')
+					b = append(b, '.')
+				}
+				b = append(b, sf.Name...)
+				b = append(b, ' ')
+			}
+			b = appendTypeName(b, sf.Type, qualified, false)
+			if sf.Tag != "" {
+				b = append(b, ' ')
+				b = strconv.AppendQuote(b, string(sf.Tag))
+			}
+		}
+		if b[len(b)-1] == ' ' {
+			b = b[:len(b)-1]
+		} else {
+			b = append(b, ' ')
+		}
+		b = append(b, '}')
+	case reflect.Slice, reflect.Array:
+		b = append(b, '[')
+		if k == reflect.Array {
+			b = strconv.AppendUint(b, uint64(t.Len()), 10)
+		}
+		b = append(b, ']')
+		b = appendTypeName(b, t.Elem(), qualified, false)
+	case reflect.Map:
+		b = append(b, "map["...)
+		b = appendTypeName(b, t.Key(), qualified, false)
+		b = append(b, ']')
+		b = appendTypeName(b, t.Elem(), qualified, false)
+	case reflect.Ptr:
+		b = append(b, '*')
+		b = appendTypeName(b, t.Elem(), qualified, false)
+	case reflect.Interface:
+		b = append(b, "interface{ "...)
+		for i := 0; i < t.NumMethod(); i++ {
+			if i > 0 {
+				b = append(b, "; "...)
+			}
+			m := t.Method(i)
+			if qualified && m.PkgPath != "" {
+				b = append(b, '"')
+				b = append(b, m.PkgPath...)
+				b = append(b, '"')
+				b = append(b, '.')
+			}
+			b = append(b, m.Name...)
+			b = appendTypeName(b, m.Type, qualified, true)
+		}
+		if b[len(b)-1] == ' ' {
+			b = b[:len(b)-1]
+		} else {
+			b = append(b, ' ')
+		}
+		b = append(b, '}')
+	default:
+		panic("invalid kind: " + k.String())
+	}
+	return b
+}
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/value/pointer_purego.go b/vendor/github.com/google/go-cmp/cmp/internal/value/pointer_purego.go
index 0a01c47..44f4a5a 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/value/pointer_purego.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/value/pointer_purego.go
@@ -1,6 +1,6 @@
 // Copyright 2018, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // +build purego
 
@@ -21,3 +21,13 @@
 	// assumes that the GC implementation does not use a moving collector.
 	return Pointer{v.Pointer(), v.Type()}
 }
+
+// IsNil reports whether the pointer is nil.
+func (p Pointer) IsNil() bool {
+	return p.p == 0
+}
+
+// Uintptr returns the pointer as a uintptr.
+func (p Pointer) Uintptr() uintptr {
+	return p.p
+}
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/value/pointer_unsafe.go b/vendor/github.com/google/go-cmp/cmp/internal/value/pointer_unsafe.go
index da134ae..a605953 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/value/pointer_unsafe.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/value/pointer_unsafe.go
@@ -1,6 +1,6 @@
 // Copyright 2018, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 // +build !purego
 
@@ -24,3 +24,13 @@
 	// which is necessary if the GC ever uses a moving collector.
 	return Pointer{unsafe.Pointer(v.Pointer()), v.Type()}
 }
+
+// IsNil reports whether the pointer is nil.
+func (p Pointer) IsNil() bool {
+	return p.p == nil
+}
+
+// Uintptr returns the pointer as a uintptr.
+func (p Pointer) Uintptr() uintptr {
+	return uintptr(p.p)
+}
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/value/sort.go b/vendor/github.com/google/go-cmp/cmp/internal/value/sort.go
index 24fbae6..98533b0 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/value/sort.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/value/sort.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package value
 
diff --git a/vendor/github.com/google/go-cmp/cmp/internal/value/zero.go b/vendor/github.com/google/go-cmp/cmp/internal/value/zero.go
index 06a8ffd..9147a29 100644
--- a/vendor/github.com/google/go-cmp/cmp/internal/value/zero.go
+++ b/vendor/github.com/google/go-cmp/cmp/internal/value/zero.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package value
 
diff --git a/vendor/github.com/google/go-cmp/cmp/options.go b/vendor/github.com/google/go-cmp/cmp/options.go
index abbd2a6..e57b9eb 100644
--- a/vendor/github.com/google/go-cmp/cmp/options.go
+++ b/vendor/github.com/google/go-cmp/cmp/options.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package cmp
 
@@ -225,11 +225,14 @@
 
 	// Unable to Interface implies unexported field without visibility access.
 	if !vx.CanInterface() || !vy.CanInterface() {
-		const help = "consider using a custom Comparer; if you control the implementation of type, you can also consider using an Exporter, AllowUnexported, or cmpopts.IgnoreUnexported"
+		help := "consider using a custom Comparer; if you control the implementation of type, you can also consider using an Exporter, AllowUnexported, or cmpopts.IgnoreUnexported"
 		var name string
 		if t := s.curPath.Index(-2).Type(); t.Name() != "" {
 			// Named type with unexported fields.
 			name = fmt.Sprintf("%q.%v", t.PkgPath(), t.Name()) // e.g., "path/to/package".MyType
+			if _, ok := reflect.New(t).Interface().(error); ok {
+				help = "consider using cmpopts.EquateErrors to compare error values"
+			}
 		} else {
 			// Unnamed type with unexported fields. Derive PkgPath from field.
 			var pkgPath string
diff --git a/vendor/github.com/google/go-cmp/cmp/path.go b/vendor/github.com/google/go-cmp/cmp/path.go
index 509d6b8..3d45c1a 100644
--- a/vendor/github.com/google/go-cmp/cmp/path.go
+++ b/vendor/github.com/google/go-cmp/cmp/path.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package cmp
 
@@ -177,7 +177,8 @@
 	// pvx, pvy, and field are only valid if unexported is true.
 	unexported bool
 	mayForce   bool                // Forcibly allow visibility
-	pvx, pvy   reflect.Value       // Parent values
+	paddr      bool                // Was parent addressable?
+	pvx, pvy   reflect.Value       // Parent values (always addressible)
 	field      reflect.StructField // Field information
 }
 
@@ -189,8 +190,8 @@
 
 	// Forcibly obtain read-write access to an unexported struct field.
 	if sf.mayForce {
-		vx = retrieveUnexportedField(sf.pvx, sf.field)
-		vy = retrieveUnexportedField(sf.pvy, sf.field)
+		vx = retrieveUnexportedField(sf.pvx, sf.field, sf.paddr)
+		vy = retrieveUnexportedField(sf.pvy, sf.field, sf.paddr)
 		return vx, vy // CanInterface reports true
 	}
 	return sf.vx, sf.vy // CanInterface reports false
diff --git a/vendor/github.com/google/go-cmp/cmp/report.go b/vendor/github.com/google/go-cmp/cmp/report.go
index 6ddf299..f43cd12 100644
--- a/vendor/github.com/google/go-cmp/cmp/report.go
+++ b/vendor/github.com/google/go-cmp/cmp/report.go
@@ -1,6 +1,6 @@
 // Copyright 2017, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package cmp
 
@@ -41,7 +41,10 @@
 	if r.root.NumDiff == 0 {
 		return ""
 	}
-	return formatOptions{}.FormatDiff(r.root).String()
+	ptrs := new(pointerReferences)
+	text := formatOptions{}.FormatDiff(r.root, ptrs)
+	resolveReferences(text)
+	return text.String()
 }
 
 func assert(ok bool) {
diff --git a/vendor/github.com/google/go-cmp/cmp/report_compare.go b/vendor/github.com/google/go-cmp/cmp/report_compare.go
index 17a05ee..104bb30 100644
--- a/vendor/github.com/google/go-cmp/cmp/report_compare.go
+++ b/vendor/github.com/google/go-cmp/cmp/report_compare.go
@@ -1,6 +1,6 @@
 // Copyright 2019, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package cmp
 
@@ -11,14 +11,6 @@
 	"github.com/google/go-cmp/cmp/internal/value"
 )
 
-// TODO: Enforce limits?
-//	* Enforce maximum number of records to print per node?
-//	* Enforce maximum size in bytes allowed?
-//	* As a heuristic, use less verbosity for equal nodes than unequal nodes.
-// TODO: Enforce unique outputs?
-//	* Avoid Stringer methods if it results in same output?
-//	* Print pointer address if outputs still equal?
-
 // numContextRecords is the number of surrounding equal records to print.
 const numContextRecords = 2
 
@@ -71,24 +63,66 @@
 	opts.TypeMode = t
 	return opts
 }
+func (opts formatOptions) WithVerbosity(level int) formatOptions {
+	opts.VerbosityLevel = level
+	opts.LimitVerbosity = true
+	return opts
+}
+func (opts formatOptions) verbosity() uint {
+	switch {
+	case opts.VerbosityLevel < 0:
+		return 0
+	case opts.VerbosityLevel > 16:
+		return 16 // some reasonable maximum to avoid shift overflow
+	default:
+		return uint(opts.VerbosityLevel)
+	}
+}
+
+const maxVerbosityPreset = 6
+
+// verbosityPreset modifies the verbosity settings given an index
+// between 0 and maxVerbosityPreset, inclusive.
+func verbosityPreset(opts formatOptions, i int) formatOptions {
+	opts.VerbosityLevel = int(opts.verbosity()) + 2*i
+	if i > 0 {
+		opts.AvoidStringer = true
+	}
+	if i >= maxVerbosityPreset {
+		opts.PrintAddresses = true
+		opts.QualifiedNames = true
+	}
+	return opts
+}
 
 // FormatDiff converts a valueNode tree into a textNode tree, where the later
 // is a textual representation of the differences detected in the former.
-func (opts formatOptions) FormatDiff(v *valueNode) textNode {
+func (opts formatOptions) FormatDiff(v *valueNode, ptrs *pointerReferences) (out textNode) {
+	if opts.DiffMode == diffIdentical {
+		opts = opts.WithVerbosity(1)
+	} else if opts.verbosity() < 3 {
+		opts = opts.WithVerbosity(3)
+	}
+
 	// Check whether we have specialized formatting for this node.
 	// This is not necessary, but helpful for producing more readable outputs.
 	if opts.CanFormatDiffSlice(v) {
 		return opts.FormatDiffSlice(v)
 	}
 
+	var parentKind reflect.Kind
+	if v.parent != nil && v.parent.TransformerName == "" {
+		parentKind = v.parent.Type.Kind()
+	}
+
 	// For leaf nodes, format the value based on the reflect.Values alone.
 	if v.MaxDepth == 0 {
 		switch opts.DiffMode {
 		case diffUnknown, diffIdentical:
 			// Format Equal.
 			if v.NumDiff == 0 {
-				outx := opts.FormatValue(v.ValueX, visitedPointers{})
-				outy := opts.FormatValue(v.ValueY, visitedPointers{})
+				outx := opts.FormatValue(v.ValueX, parentKind, ptrs)
+				outy := opts.FormatValue(v.ValueY, parentKind, ptrs)
 				if v.NumIgnored > 0 && v.NumSame == 0 {
 					return textEllipsis
 				} else if outx.Len() < outy.Len() {
@@ -101,8 +135,13 @@
 			// Format unequal.
 			assert(opts.DiffMode == diffUnknown)
 			var list textList
-			outx := opts.WithTypeMode(elideType).FormatValue(v.ValueX, visitedPointers{})
-			outy := opts.WithTypeMode(elideType).FormatValue(v.ValueY, visitedPointers{})
+			outx := opts.WithTypeMode(elideType).FormatValue(v.ValueX, parentKind, ptrs)
+			outy := opts.WithTypeMode(elideType).FormatValue(v.ValueY, parentKind, ptrs)
+			for i := 0; i <= maxVerbosityPreset && outx != nil && outy != nil && outx.Equal(outy); i++ {
+				opts2 := verbosityPreset(opts, i).WithTypeMode(elideType)
+				outx = opts2.FormatValue(v.ValueX, parentKind, ptrs)
+				outy = opts2.FormatValue(v.ValueY, parentKind, ptrs)
+			}
 			if outx != nil {
 				list = append(list, textRecord{Diff: '-', Value: outx})
 			}
@@ -111,34 +150,57 @@
 			}
 			return opts.WithTypeMode(emitType).FormatType(v.Type, list)
 		case diffRemoved:
-			return opts.FormatValue(v.ValueX, visitedPointers{})
+			return opts.FormatValue(v.ValueX, parentKind, ptrs)
 		case diffInserted:
-			return opts.FormatValue(v.ValueY, visitedPointers{})
+			return opts.FormatValue(v.ValueY, parentKind, ptrs)
 		default:
 			panic("invalid diff mode")
 		}
 	}
 
+	// Register slice element to support cycle detection.
+	if parentKind == reflect.Slice {
+		ptrRefs := ptrs.PushPair(v.ValueX, v.ValueY, opts.DiffMode, true)
+		defer ptrs.Pop()
+		defer func() { out = wrapTrunkReferences(ptrRefs, out) }()
+	}
+
 	// Descend into the child value node.
 	if v.TransformerName != "" {
-		out := opts.WithTypeMode(emitType).FormatDiff(v.Value)
-		out = textWrap{"Inverse(" + v.TransformerName + ", ", out, ")"}
+		out := opts.WithTypeMode(emitType).FormatDiff(v.Value, ptrs)
+		out = &textWrap{Prefix: "Inverse(" + v.TransformerName + ", ", Value: out, Suffix: ")"}
 		return opts.FormatType(v.Type, out)
 	} else {
 		switch k := v.Type.Kind(); k {
-		case reflect.Struct, reflect.Array, reflect.Slice, reflect.Map:
-			return opts.FormatType(v.Type, opts.formatDiffList(v.Records, k))
+		case reflect.Struct, reflect.Array, reflect.Slice:
+			out = opts.formatDiffList(v.Records, k, ptrs)
+			out = opts.FormatType(v.Type, out)
+		case reflect.Map:
+			// Register map to support cycle detection.
+			ptrRefs := ptrs.PushPair(v.ValueX, v.ValueY, opts.DiffMode, false)
+			defer ptrs.Pop()
+
+			out = opts.formatDiffList(v.Records, k, ptrs)
+			out = wrapTrunkReferences(ptrRefs, out)
+			out = opts.FormatType(v.Type, out)
 		case reflect.Ptr:
-			return textWrap{"&", opts.FormatDiff(v.Value), ""}
+			// Register pointer to support cycle detection.
+			ptrRefs := ptrs.PushPair(v.ValueX, v.ValueY, opts.DiffMode, false)
+			defer ptrs.Pop()
+
+			out = opts.FormatDiff(v.Value, ptrs)
+			out = wrapTrunkReferences(ptrRefs, out)
+			out = &textWrap{Prefix: "&", Value: out}
 		case reflect.Interface:
-			return opts.WithTypeMode(emitType).FormatDiff(v.Value)
+			out = opts.WithTypeMode(emitType).FormatDiff(v.Value, ptrs)
 		default:
 			panic(fmt.Sprintf("%v cannot have children", k))
 		}
+		return out
 	}
 }
 
-func (opts formatOptions) formatDiffList(recs []reportRecord, k reflect.Kind) textNode {
+func (opts formatOptions) formatDiffList(recs []reportRecord, k reflect.Kind, ptrs *pointerReferences) textNode {
 	// Derive record name based on the data structure kind.
 	var name string
 	var formatKey func(reflect.Value) string
@@ -154,7 +216,17 @@
 	case reflect.Map:
 		name = "entry"
 		opts = opts.WithTypeMode(elideType)
-		formatKey = formatMapKey
+		formatKey = func(v reflect.Value) string { return formatMapKey(v, false, ptrs) }
+	}
+
+	maxLen := -1
+	if opts.LimitVerbosity {
+		if opts.DiffMode == diffIdentical {
+			maxLen = ((1 << opts.verbosity()) >> 1) << 2 // 0, 4, 8, 16, 32, etc...
+		} else {
+			maxLen = (1 << opts.verbosity()) << 1 // 2, 4, 8, 16, 32, 64, etc...
+		}
+		opts.VerbosityLevel--
 	}
 
 	// Handle unification.
@@ -163,6 +235,11 @@
 		var list textList
 		var deferredEllipsis bool // Add final "..." to indicate records were dropped
 		for _, r := range recs {
+			if len(list) == maxLen {
+				deferredEllipsis = true
+				break
+			}
+
 			// Elide struct fields that are zero value.
 			if k == reflect.Struct {
 				var isZero bool
@@ -186,23 +263,31 @@
 				}
 				continue
 			}
-			if out := opts.FormatDiff(r.Value); out != nil {
+			if out := opts.FormatDiff(r.Value, ptrs); out != nil {
 				list = append(list, textRecord{Key: formatKey(r.Key), Value: out})
 			}
 		}
 		if deferredEllipsis {
 			list.AppendEllipsis(diffStats{})
 		}
-		return textWrap{"{", list, "}"}
+		return &textWrap{Prefix: "{", Value: list, Suffix: "}"}
 	case diffUnknown:
 	default:
 		panic("invalid diff mode")
 	}
 
 	// Handle differencing.
+	var numDiffs int
 	var list textList
+	var keys []reflect.Value // invariant: len(list) == len(keys)
 	groups := coalesceAdjacentRecords(name, recs)
+	maxGroup := diffStats{Name: name}
 	for i, ds := range groups {
+		if maxLen >= 0 && numDiffs >= maxLen {
+			maxGroup = maxGroup.Append(ds)
+			continue
+		}
+
 		// Handle equal records.
 		if ds.NumDiff() == 0 {
 			// Compute the number of leading and trailing records to print.
@@ -226,16 +311,21 @@
 
 			// Format the equal values.
 			for _, r := range recs[:numLo] {
-				out := opts.WithDiffMode(diffIdentical).FormatDiff(r.Value)
+				out := opts.WithDiffMode(diffIdentical).FormatDiff(r.Value, ptrs)
 				list = append(list, textRecord{Key: formatKey(r.Key), Value: out})
+				keys = append(keys, r.Key)
 			}
 			if numEqual > numLo+numHi {
 				ds.NumIdentical -= numLo + numHi
 				list.AppendEllipsis(ds)
+				for len(keys) < len(list) {
+					keys = append(keys, reflect.Value{})
+				}
 			}
 			for _, r := range recs[numEqual-numHi : numEqual] {
-				out := opts.WithDiffMode(diffIdentical).FormatDiff(r.Value)
+				out := opts.WithDiffMode(diffIdentical).FormatDiff(r.Value, ptrs)
 				list = append(list, textRecord{Key: formatKey(r.Key), Value: out})
+				keys = append(keys, r.Key)
 			}
 			recs = recs[numEqual:]
 			continue
@@ -247,24 +337,70 @@
 			case opts.CanFormatDiffSlice(r.Value):
 				out := opts.FormatDiffSlice(r.Value)
 				list = append(list, textRecord{Key: formatKey(r.Key), Value: out})
+				keys = append(keys, r.Key)
 			case r.Value.NumChildren == r.Value.MaxDepth:
-				outx := opts.WithDiffMode(diffRemoved).FormatDiff(r.Value)
-				outy := opts.WithDiffMode(diffInserted).FormatDiff(r.Value)
+				outx := opts.WithDiffMode(diffRemoved).FormatDiff(r.Value, ptrs)
+				outy := opts.WithDiffMode(diffInserted).FormatDiff(r.Value, ptrs)
+				for i := 0; i <= maxVerbosityPreset && outx != nil && outy != nil && outx.Equal(outy); i++ {
+					opts2 := verbosityPreset(opts, i)
+					outx = opts2.WithDiffMode(diffRemoved).FormatDiff(r.Value, ptrs)
+					outy = opts2.WithDiffMode(diffInserted).FormatDiff(r.Value, ptrs)
+				}
 				if outx != nil {
 					list = append(list, textRecord{Diff: diffRemoved, Key: formatKey(r.Key), Value: outx})
+					keys = append(keys, r.Key)
 				}
 				if outy != nil {
 					list = append(list, textRecord{Diff: diffInserted, Key: formatKey(r.Key), Value: outy})
+					keys = append(keys, r.Key)
 				}
 			default:
-				out := opts.FormatDiff(r.Value)
+				out := opts.FormatDiff(r.Value, ptrs)
 				list = append(list, textRecord{Key: formatKey(r.Key), Value: out})
+				keys = append(keys, r.Key)
 			}
 		}
 		recs = recs[ds.NumDiff():]
+		numDiffs += ds.NumDiff()
 	}
-	assert(len(recs) == 0)
-	return textWrap{"{", list, "}"}
+	if maxGroup.IsZero() {
+		assert(len(recs) == 0)
+	} else {
+		list.AppendEllipsis(maxGroup)
+		for len(keys) < len(list) {
+			keys = append(keys, reflect.Value{})
+		}
+	}
+	assert(len(list) == len(keys))
+
+	// For maps, the default formatting logic uses fmt.Stringer which may
+	// produce ambiguous output. Avoid calling String to disambiguate.
+	if k == reflect.Map {
+		var ambiguous bool
+		seenKeys := map[string]reflect.Value{}
+		for i, currKey := range keys {
+			if currKey.IsValid() {
+				strKey := list[i].Key
+				prevKey, seen := seenKeys[strKey]
+				if seen && prevKey.CanInterface() && currKey.CanInterface() {
+					ambiguous = prevKey.Interface() != currKey.Interface()
+					if ambiguous {
+						break
+					}
+				}
+				seenKeys[strKey] = currKey
+			}
+		}
+		if ambiguous {
+			for i, k := range keys {
+				if k.IsValid() {
+					list[i].Key = formatMapKey(k, true, ptrs)
+				}
+			}
+		}
+	}
+
+	return &textWrap{Prefix: "{", Value: list, Suffix: "}"}
 }
 
 // coalesceAdjacentRecords coalesces the list of records into groups of
diff --git a/vendor/github.com/google/go-cmp/cmp/report_references.go b/vendor/github.com/google/go-cmp/cmp/report_references.go
new file mode 100644
index 0000000..be31b33
--- /dev/null
+++ b/vendor/github.com/google/go-cmp/cmp/report_references.go
@@ -0,0 +1,264 @@
+// Copyright 2020, The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package cmp
+
+import (
+	"fmt"
+	"reflect"
+	"strings"
+
+	"github.com/google/go-cmp/cmp/internal/flags"
+	"github.com/google/go-cmp/cmp/internal/value"
+)
+
+const (
+	pointerDelimPrefix = "⟪"
+	pointerDelimSuffix = "⟫"
+)
+
+// formatPointer prints the address of the pointer.
+func formatPointer(p value.Pointer, withDelims bool) string {
+	v := p.Uintptr()
+	if flags.Deterministic {
+		v = 0xdeadf00f // Only used for stable testing purposes
+	}
+	if withDelims {
+		return pointerDelimPrefix + formatHex(uint64(v)) + pointerDelimSuffix
+	}
+	return formatHex(uint64(v))
+}
+
+// pointerReferences is a stack of pointers visited so far.
+type pointerReferences [][2]value.Pointer
+
+func (ps *pointerReferences) PushPair(vx, vy reflect.Value, d diffMode, deref bool) (pp [2]value.Pointer) {
+	if deref && vx.IsValid() {
+		vx = vx.Addr()
+	}
+	if deref && vy.IsValid() {
+		vy = vy.Addr()
+	}
+	switch d {
+	case diffUnknown, diffIdentical:
+		pp = [2]value.Pointer{value.PointerOf(vx), value.PointerOf(vy)}
+	case diffRemoved:
+		pp = [2]value.Pointer{value.PointerOf(vx), value.Pointer{}}
+	case diffInserted:
+		pp = [2]value.Pointer{value.Pointer{}, value.PointerOf(vy)}
+	}
+	*ps = append(*ps, pp)
+	return pp
+}
+
+func (ps *pointerReferences) Push(v reflect.Value) (p value.Pointer, seen bool) {
+	p = value.PointerOf(v)
+	for _, pp := range *ps {
+		if p == pp[0] || p == pp[1] {
+			return p, true
+		}
+	}
+	*ps = append(*ps, [2]value.Pointer{p, p})
+	return p, false
+}
+
+func (ps *pointerReferences) Pop() {
+	*ps = (*ps)[:len(*ps)-1]
+}
+
+// trunkReferences is metadata for a textNode indicating that the sub-tree
+// represents the value for either pointer in a pair of references.
+type trunkReferences struct{ pp [2]value.Pointer }
+
+// trunkReference is metadata for a textNode indicating that the sub-tree
+// represents the value for the given pointer reference.
+type trunkReference struct{ p value.Pointer }
+
+// leafReference is metadata for a textNode indicating that the value is
+// truncated as it refers to another part of the tree (i.e., a trunk).
+type leafReference struct{ p value.Pointer }
+
+func wrapTrunkReferences(pp [2]value.Pointer, s textNode) textNode {
+	switch {
+	case pp[0].IsNil():
+		return &textWrap{Value: s, Metadata: trunkReference{pp[1]}}
+	case pp[1].IsNil():
+		return &textWrap{Value: s, Metadata: trunkReference{pp[0]}}
+	case pp[0] == pp[1]:
+		return &textWrap{Value: s, Metadata: trunkReference{pp[0]}}
+	default:
+		return &textWrap{Value: s, Metadata: trunkReferences{pp}}
+	}
+}
+func wrapTrunkReference(p value.Pointer, printAddress bool, s textNode) textNode {
+	var prefix string
+	if printAddress {
+		prefix = formatPointer(p, true)
+	}
+	return &textWrap{Prefix: prefix, Value: s, Metadata: trunkReference{p}}
+}
+func makeLeafReference(p value.Pointer, printAddress bool) textNode {
+	out := &textWrap{Prefix: "(", Value: textEllipsis, Suffix: ")"}
+	var prefix string
+	if printAddress {
+		prefix = formatPointer(p, true)
+	}
+	return &textWrap{Prefix: prefix, Value: out, Metadata: leafReference{p}}
+}
+
+// resolveReferences walks the textNode tree searching for any leaf reference
+// metadata and resolves each against the corresponding trunk references.
+// Since pointer addresses in memory are not particularly readable to the user,
+// it replaces each pointer value with an arbitrary and unique reference ID.
+func resolveReferences(s textNode) {
+	var walkNodes func(textNode, func(textNode))
+	walkNodes = func(s textNode, f func(textNode)) {
+		f(s)
+		switch s := s.(type) {
+		case *textWrap:
+			walkNodes(s.Value, f)
+		case textList:
+			for _, r := range s {
+				walkNodes(r.Value, f)
+			}
+		}
+	}
+
+	// Collect all trunks and leaves with reference metadata.
+	var trunks, leaves []*textWrap
+	walkNodes(s, func(s textNode) {
+		if s, ok := s.(*textWrap); ok {
+			switch s.Metadata.(type) {
+			case leafReference:
+				leaves = append(leaves, s)
+			case trunkReference, trunkReferences:
+				trunks = append(trunks, s)
+			}
+		}
+	})
+
+	// No leaf references to resolve.
+	if len(leaves) == 0 {
+		return
+	}
+
+	// Collect the set of all leaf references to resolve.
+	leafPtrs := make(map[value.Pointer]bool)
+	for _, leaf := range leaves {
+		leafPtrs[leaf.Metadata.(leafReference).p] = true
+	}
+
+	// Collect the set of trunk pointers that are always paired together.
+	// This allows us to assign a single ID to both pointers for brevity.
+	// If a pointer in a pair ever occurs by itself or as a different pair,
+	// then the pair is broken.
+	pairedTrunkPtrs := make(map[value.Pointer]value.Pointer)
+	unpair := func(p value.Pointer) {
+		if !pairedTrunkPtrs[p].IsNil() {
+			pairedTrunkPtrs[pairedTrunkPtrs[p]] = value.Pointer{} // invalidate other half
+		}
+		pairedTrunkPtrs[p] = value.Pointer{} // invalidate this half
+	}
+	for _, trunk := range trunks {
+		switch p := trunk.Metadata.(type) {
+		case trunkReference:
+			unpair(p.p) // standalone pointer cannot be part of a pair
+		case trunkReferences:
+			p0, ok0 := pairedTrunkPtrs[p.pp[0]]
+			p1, ok1 := pairedTrunkPtrs[p.pp[1]]
+			switch {
+			case !ok0 && !ok1:
+				// Register the newly seen pair.
+				pairedTrunkPtrs[p.pp[0]] = p.pp[1]
+				pairedTrunkPtrs[p.pp[1]] = p.pp[0]
+			case ok0 && ok1 && p0 == p.pp[1] && p1 == p.pp[0]:
+				// Exact pair already seen; do nothing.
+			default:
+				// Pair conflicts with some other pair; break all pairs.
+				unpair(p.pp[0])
+				unpair(p.pp[1])
+			}
+		}
+	}
+
+	// Correlate each pointer referenced by leaves to a unique identifier,
+	// and print the IDs for each trunk that matches those pointers.
+	var nextID uint
+	ptrIDs := make(map[value.Pointer]uint)
+	newID := func() uint {
+		id := nextID
+		nextID++
+		return id
+	}
+	for _, trunk := range trunks {
+		switch p := trunk.Metadata.(type) {
+		case trunkReference:
+			if print := leafPtrs[p.p]; print {
+				id, ok := ptrIDs[p.p]
+				if !ok {
+					id = newID()
+					ptrIDs[p.p] = id
+				}
+				trunk.Prefix = updateReferencePrefix(trunk.Prefix, formatReference(id))
+			}
+		case trunkReferences:
+			print0 := leafPtrs[p.pp[0]]
+			print1 := leafPtrs[p.pp[1]]
+			if print0 || print1 {
+				id0, ok0 := ptrIDs[p.pp[0]]
+				id1, ok1 := ptrIDs[p.pp[1]]
+				isPair := pairedTrunkPtrs[p.pp[0]] == p.pp[1] && pairedTrunkPtrs[p.pp[1]] == p.pp[0]
+				if isPair {
+					var id uint
+					assert(ok0 == ok1) // must be seen together or not at all
+					if ok0 {
+						assert(id0 == id1) // must have the same ID
+						id = id0
+					} else {
+						id = newID()
+						ptrIDs[p.pp[0]] = id
+						ptrIDs[p.pp[1]] = id
+					}
+					trunk.Prefix = updateReferencePrefix(trunk.Prefix, formatReference(id))
+				} else {
+					if print0 && !ok0 {
+						id0 = newID()
+						ptrIDs[p.pp[0]] = id0
+					}
+					if print1 && !ok1 {
+						id1 = newID()
+						ptrIDs[p.pp[1]] = id1
+					}
+					switch {
+					case print0 && print1:
+						trunk.Prefix = updateReferencePrefix(trunk.Prefix, formatReference(id0)+","+formatReference(id1))
+					case print0:
+						trunk.Prefix = updateReferencePrefix(trunk.Prefix, formatReference(id0))
+					case print1:
+						trunk.Prefix = updateReferencePrefix(trunk.Prefix, formatReference(id1))
+					}
+				}
+			}
+		}
+	}
+
+	// Update all leaf references with the unique identifier.
+	for _, leaf := range leaves {
+		if id, ok := ptrIDs[leaf.Metadata.(leafReference).p]; ok {
+			leaf.Prefix = updateReferencePrefix(leaf.Prefix, formatReference(id))
+		}
+	}
+}
+
+func formatReference(id uint) string {
+	return fmt.Sprintf("ref#%d", id)
+}
+
+func updateReferencePrefix(prefix, ref string) string {
+	if prefix == "" {
+		return pointerDelimPrefix + ref + pointerDelimSuffix
+	}
+	suffix := strings.TrimPrefix(prefix, pointerDelimPrefix)
+	return pointerDelimPrefix + ref + ": " + suffix
+}
diff --git a/vendor/github.com/google/go-cmp/cmp/report_reflect.go b/vendor/github.com/google/go-cmp/cmp/report_reflect.go
index 2761b62..33f0357 100644
--- a/vendor/github.com/google/go-cmp/cmp/report_reflect.go
+++ b/vendor/github.com/google/go-cmp/cmp/report_reflect.go
@@ -1,17 +1,18 @@
 // Copyright 2019, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package cmp
 
 import (
+	"bytes"
 	"fmt"
 	"reflect"
 	"strconv"
 	"strings"
 	"unicode"
+	"unicode/utf8"
 
-	"github.com/google/go-cmp/cmp/internal/flags"
 	"github.com/google/go-cmp/cmp/internal/value"
 )
 
@@ -20,14 +21,22 @@
 	// methods like error.Error or fmt.Stringer.String.
 	AvoidStringer bool
 
-	// ShallowPointers controls whether to avoid descending into pointers.
-	// Useful when printing map keys, where pointer comparison is performed
-	// on the pointer address rather than the pointed-at value.
-	ShallowPointers bool
-
 	// PrintAddresses controls whether to print the address of all pointers,
 	// slice elements, and maps.
 	PrintAddresses bool
+
+	// QualifiedNames controls whether FormatType uses the fully qualified name
+	// (including the full package path as opposed to just the package name).
+	QualifiedNames bool
+
+	// VerbosityLevel controls the amount of output to produce.
+	// A higher value produces more output. A value of zero or lower produces
+	// no output (represented using an ellipsis).
+	// If LimitVerbosity is false, then the level is treated as infinite.
+	VerbosityLevel int
+
+	// LimitVerbosity specifies that formatting should respect VerbosityLevel.
+	LimitVerbosity bool
 }
 
 // FormatType prints the type as if it were wrapping s.
@@ -44,12 +53,15 @@
 		default:
 			return s
 		}
+		if opts.DiffMode == diffIdentical {
+			return s // elide type for identical nodes
+		}
 	case elideType:
 		return s
 	}
 
 	// Determine the type label, applying special handling for unnamed types.
-	typeName := t.String()
+	typeName := value.TypeString(t, opts.QualifiedNames)
 	if t.Name() == "" {
 		// According to Go grammar, certain type literals contain symbols that
 		// do not strongly bind to the next lexicographical token (e.g., *T).
@@ -57,39 +69,77 @@
 		case reflect.Chan, reflect.Func, reflect.Ptr:
 			typeName = "(" + typeName + ")"
 		}
-		typeName = strings.Replace(typeName, "struct {", "struct{", -1)
-		typeName = strings.Replace(typeName, "interface {", "interface{", -1)
 	}
+	return &textWrap{Prefix: typeName, Value: wrapParens(s)}
+}
 
-	// Avoid wrap the value in parenthesis if unnecessary.
-	if s, ok := s.(textWrap); ok {
-		hasParens := strings.HasPrefix(s.Prefix, "(") && strings.HasSuffix(s.Suffix, ")")
-		hasBraces := strings.HasPrefix(s.Prefix, "{") && strings.HasSuffix(s.Suffix, "}")
+// wrapParens wraps s with a set of parenthesis, but avoids it if the
+// wrapped node itself is already surrounded by a pair of parenthesis or braces.
+// It handles unwrapping one level of pointer-reference nodes.
+func wrapParens(s textNode) textNode {
+	var refNode *textWrap
+	if s2, ok := s.(*textWrap); ok {
+		// Unwrap a single pointer reference node.
+		switch s2.Metadata.(type) {
+		case leafReference, trunkReference, trunkReferences:
+			refNode = s2
+			if s3, ok := refNode.Value.(*textWrap); ok {
+				s2 = s3
+			}
+		}
+
+		// Already has delimiters that make parenthesis unnecessary.
+		hasParens := strings.HasPrefix(s2.Prefix, "(") && strings.HasSuffix(s2.Suffix, ")")
+		hasBraces := strings.HasPrefix(s2.Prefix, "{") && strings.HasSuffix(s2.Suffix, "}")
 		if hasParens || hasBraces {
-			return textWrap{typeName, s, ""}
+			return s
 		}
 	}
-	return textWrap{typeName + "(", s, ")"}
+	if refNode != nil {
+		refNode.Value = &textWrap{Prefix: "(", Value: refNode.Value, Suffix: ")"}
+		return s
+	}
+	return &textWrap{Prefix: "(", Value: s, Suffix: ")"}
 }
 
 // FormatValue prints the reflect.Value, taking extra care to avoid descending
-// into pointers already in m. As pointers are visited, m is also updated.
-func (opts formatOptions) FormatValue(v reflect.Value, m visitedPointers) (out textNode) {
+// into pointers already in ptrs. As pointers are visited, ptrs is also updated.
+func (opts formatOptions) FormatValue(v reflect.Value, parentKind reflect.Kind, ptrs *pointerReferences) (out textNode) {
 	if !v.IsValid() {
 		return nil
 	}
 	t := v.Type()
 
+	// Check slice element for cycles.
+	if parentKind == reflect.Slice {
+		ptrRef, visited := ptrs.Push(v.Addr())
+		if visited {
+			return makeLeafReference(ptrRef, false)
+		}
+		defer ptrs.Pop()
+		defer func() { out = wrapTrunkReference(ptrRef, false, out) }()
+	}
+
 	// Check whether there is an Error or String method to call.
 	if !opts.AvoidStringer && v.CanInterface() {
 		// Avoid calling Error or String methods on nil receivers since many
 		// implementations crash when doing so.
 		if (t.Kind() != reflect.Ptr && t.Kind() != reflect.Interface) || !v.IsNil() {
-			switch v := v.Interface().(type) {
-			case error:
-				return textLine("e" + formatString(v.Error()))
-			case fmt.Stringer:
-				return textLine("s" + formatString(v.String()))
+			var prefix, strVal string
+			func() {
+				// Swallow and ignore any panics from String or Error.
+				defer func() { recover() }()
+				switch v := v.Interface().(type) {
+				case error:
+					strVal = v.Error()
+					prefix = "e"
+				case fmt.Stringer:
+					strVal = v.String()
+					prefix = "s"
+				}
+			}()
+			if prefix != "" {
+				return opts.formatString(prefix, strVal)
 			}
 		}
 	}
@@ -102,94 +152,140 @@
 		}
 	}()
 
-	var ptr string
 	switch t.Kind() {
 	case reflect.Bool:
 		return textLine(fmt.Sprint(v.Bool()))
 	case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
 		return textLine(fmt.Sprint(v.Int()))
-	case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
-		// Unnamed uints are usually bytes or words, so use hexadecimal.
-		if t.PkgPath() == "" || t.Kind() == reflect.Uintptr {
+	case reflect.Uint, reflect.Uint16, reflect.Uint32, reflect.Uint64:
+		return textLine(fmt.Sprint(v.Uint()))
+	case reflect.Uint8:
+		if parentKind == reflect.Slice || parentKind == reflect.Array {
 			return textLine(formatHex(v.Uint()))
 		}
 		return textLine(fmt.Sprint(v.Uint()))
+	case reflect.Uintptr:
+		return textLine(formatHex(v.Uint()))
 	case reflect.Float32, reflect.Float64:
 		return textLine(fmt.Sprint(v.Float()))
 	case reflect.Complex64, reflect.Complex128:
 		return textLine(fmt.Sprint(v.Complex()))
 	case reflect.String:
-		return textLine(formatString(v.String()))
+		return opts.formatString("", v.String())
 	case reflect.UnsafePointer, reflect.Chan, reflect.Func:
-		return textLine(formatPointer(v))
+		return textLine(formatPointer(value.PointerOf(v), true))
 	case reflect.Struct:
 		var list textList
+		v := makeAddressable(v) // needed for retrieveUnexportedField
+		maxLen := v.NumField()
+		if opts.LimitVerbosity {
+			maxLen = ((1 << opts.verbosity()) >> 1) << 2 // 0, 4, 8, 16, 32, etc...
+			opts.VerbosityLevel--
+		}
 		for i := 0; i < v.NumField(); i++ {
 			vv := v.Field(i)
 			if value.IsZero(vv) {
 				continue // Elide fields with zero values
 			}
-			s := opts.WithTypeMode(autoType).FormatValue(vv, m)
-			list = append(list, textRecord{Key: t.Field(i).Name, Value: s})
+			if len(list) == maxLen {
+				list.AppendEllipsis(diffStats{})
+				break
+			}
+			sf := t.Field(i)
+			if supportExporters && !isExported(sf.Name) {
+				vv = retrieveUnexportedField(v, sf, true)
+			}
+			s := opts.WithTypeMode(autoType).FormatValue(vv, t.Kind(), ptrs)
+			list = append(list, textRecord{Key: sf.Name, Value: s})
 		}
-		return textWrap{"{", list, "}"}
+		return &textWrap{Prefix: "{", Value: list, Suffix: "}"}
 	case reflect.Slice:
 		if v.IsNil() {
 			return textNil
 		}
-		if opts.PrintAddresses {
-			ptr = formatPointer(v)
+
+		// Check whether this is a []byte of text data.
+		if t.Elem() == reflect.TypeOf(byte(0)) {
+			b := v.Bytes()
+			isPrintSpace := func(r rune) bool { return unicode.IsPrint(r) && unicode.IsSpace(r) }
+			if len(b) > 0 && utf8.Valid(b) && len(bytes.TrimFunc(b, isPrintSpace)) == 0 {
+				out = opts.formatString("", string(b))
+				return opts.WithTypeMode(emitType).FormatType(t, out)
+			}
 		}
+
 		fallthrough
 	case reflect.Array:
+		maxLen := v.Len()
+		if opts.LimitVerbosity {
+			maxLen = ((1 << opts.verbosity()) >> 1) << 2 // 0, 4, 8, 16, 32, etc...
+			opts.VerbosityLevel--
+		}
 		var list textList
 		for i := 0; i < v.Len(); i++ {
-			vi := v.Index(i)
-			if vi.CanAddr() { // Check for cyclic elements
-				p := vi.Addr()
-				if m.Visit(p) {
-					var out textNode
-					out = textLine(formatPointer(p))
-					out = opts.WithTypeMode(emitType).FormatType(p.Type(), out)
-					out = textWrap{"*", out, ""}
-					list = append(list, textRecord{Value: out})
-					continue
-				}
+			if len(list) == maxLen {
+				list.AppendEllipsis(diffStats{})
+				break
 			}
-			s := opts.WithTypeMode(elideType).FormatValue(vi, m)
+			s := opts.WithTypeMode(elideType).FormatValue(v.Index(i), t.Kind(), ptrs)
 			list = append(list, textRecord{Value: s})
 		}
-		return textWrap{ptr + "{", list, "}"}
+
+		out = &textWrap{Prefix: "{", Value: list, Suffix: "}"}
+		if t.Kind() == reflect.Slice && opts.PrintAddresses {
+			header := fmt.Sprintf("ptr:%v, len:%d, cap:%d", formatPointer(value.PointerOf(v), false), v.Len(), v.Cap())
+			out = &textWrap{Prefix: pointerDelimPrefix + header + pointerDelimSuffix, Value: out}
+		}
+		return out
 	case reflect.Map:
 		if v.IsNil() {
 			return textNil
 		}
-		if m.Visit(v) {
-			return textLine(formatPointer(v))
-		}
 
+		// Check pointer for cycles.
+		ptrRef, visited := ptrs.Push(v)
+		if visited {
+			return makeLeafReference(ptrRef, opts.PrintAddresses)
+		}
+		defer ptrs.Pop()
+
+		maxLen := v.Len()
+		if opts.LimitVerbosity {
+			maxLen = ((1 << opts.verbosity()) >> 1) << 2 // 0, 4, 8, 16, 32, etc...
+			opts.VerbosityLevel--
+		}
 		var list textList
 		for _, k := range value.SortKeys(v.MapKeys()) {
-			sk := formatMapKey(k)
-			sv := opts.WithTypeMode(elideType).FormatValue(v.MapIndex(k), m)
+			if len(list) == maxLen {
+				list.AppendEllipsis(diffStats{})
+				break
+			}
+			sk := formatMapKey(k, false, ptrs)
+			sv := opts.WithTypeMode(elideType).FormatValue(v.MapIndex(k), t.Kind(), ptrs)
 			list = append(list, textRecord{Key: sk, Value: sv})
 		}
-		if opts.PrintAddresses {
-			ptr = formatPointer(v)
-		}
-		return textWrap{ptr + "{", list, "}"}
+
+		out = &textWrap{Prefix: "{", Value: list, Suffix: "}"}
+		out = wrapTrunkReference(ptrRef, opts.PrintAddresses, out)
+		return out
 	case reflect.Ptr:
 		if v.IsNil() {
 			return textNil
 		}
-		if m.Visit(v) || opts.ShallowPointers {
-			return textLine(formatPointer(v))
+
+		// Check pointer for cycles.
+		ptrRef, visited := ptrs.Push(v)
+		if visited {
+			out = makeLeafReference(ptrRef, opts.PrintAddresses)
+			return &textWrap{Prefix: "&", Value: out}
 		}
-		if opts.PrintAddresses {
-			ptr = formatPointer(v)
-		}
+		defer ptrs.Pop()
+
 		skipType = true // Let the underlying value print the type instead
-		return textWrap{"&" + ptr, opts.FormatValue(v.Elem(), m), ""}
+		out = opts.FormatValue(v.Elem(), t.Kind(), ptrs)
+		out = wrapTrunkReference(ptrRef, opts.PrintAddresses, out)
+		out = &textWrap{Prefix: "&", Value: out}
+		return out
 	case reflect.Interface:
 		if v.IsNil() {
 			return textNil
@@ -197,19 +293,67 @@
 		// Interfaces accept different concrete types,
 		// so configure the underlying value to explicitly print the type.
 		skipType = true // Print the concrete type instead
-		return opts.WithTypeMode(emitType).FormatValue(v.Elem(), m)
+		return opts.WithTypeMode(emitType).FormatValue(v.Elem(), t.Kind(), ptrs)
 	default:
 		panic(fmt.Sprintf("%v kind not handled", v.Kind()))
 	}
 }
 
+func (opts formatOptions) formatString(prefix, s string) textNode {
+	maxLen := len(s)
+	maxLines := strings.Count(s, "\n") + 1
+	if opts.LimitVerbosity {
+		maxLen = (1 << opts.verbosity()) << 5   // 32, 64, 128, 256, etc...
+		maxLines = (1 << opts.verbosity()) << 2 //  4, 8, 16, 32, 64, etc...
+	}
+
+	// For multiline strings, use the triple-quote syntax,
+	// but only use it when printing removed or inserted nodes since
+	// we only want the extra verbosity for those cases.
+	lines := strings.Split(strings.TrimSuffix(s, "\n"), "\n")
+	isTripleQuoted := len(lines) >= 4 && (opts.DiffMode == '-' || opts.DiffMode == '+')
+	for i := 0; i < len(lines) && isTripleQuoted; i++ {
+		lines[i] = strings.TrimPrefix(strings.TrimSuffix(lines[i], "\r"), "\r") // trim leading/trailing carriage returns for legacy Windows endline support
+		isPrintable := func(r rune) bool {
+			return unicode.IsPrint(r) || r == '\t' // specially treat tab as printable
+		}
+		line := lines[i]
+		isTripleQuoted = !strings.HasPrefix(strings.TrimPrefix(line, prefix), `"""`) && !strings.HasPrefix(line, "...") && strings.TrimFunc(line, isPrintable) == "" && len(line) <= maxLen
+	}
+	if isTripleQuoted {
+		var list textList
+		list = append(list, textRecord{Diff: opts.DiffMode, Value: textLine(prefix + `"""`), ElideComma: true})
+		for i, line := range lines {
+			if numElided := len(lines) - i; i == maxLines-1 && numElided > 1 {
+				comment := commentString(fmt.Sprintf("%d elided lines", numElided))
+				list = append(list, textRecord{Diff: opts.DiffMode, Value: textEllipsis, ElideComma: true, Comment: comment})
+				break
+			}
+			list = append(list, textRecord{Diff: opts.DiffMode, Value: textLine(line), ElideComma: true})
+		}
+		list = append(list, textRecord{Diff: opts.DiffMode, Value: textLine(prefix + `"""`), ElideComma: true})
+		return &textWrap{Prefix: "(", Value: list, Suffix: ")"}
+	}
+
+	// Format the string as a single-line quoted string.
+	if len(s) > maxLen+len(textEllipsis) {
+		return textLine(prefix + formatString(s[:maxLen]) + string(textEllipsis))
+	}
+	return textLine(prefix + formatString(s))
+}
+
 // formatMapKey formats v as if it were a map key.
 // The result is guaranteed to be a single line.
-func formatMapKey(v reflect.Value) string {
+func formatMapKey(v reflect.Value, disambiguate bool, ptrs *pointerReferences) string {
 	var opts formatOptions
+	opts.DiffMode = diffIdentical
 	opts.TypeMode = elideType
-	opts.ShallowPointers = true
-	s := opts.FormatValue(v, visitedPointers{}).String()
+	opts.PrintAddresses = disambiguate
+	opts.AvoidStringer = disambiguate
+	opts.QualifiedNames = disambiguate
+	opts.VerbosityLevel = maxVerbosityPreset
+	opts.LimitVerbosity = true
+	s := opts.FormatValue(v, reflect.Map, ptrs).String()
 	return strings.TrimSpace(s)
 }
 
@@ -227,7 +371,7 @@
 	rawInvalid := func(r rune) bool {
 		return r == '`' || r == '\n' || !(unicode.IsPrint(r) || r == '\t')
 	}
-	if strings.IndexFunc(s, rawInvalid) < 0 {
+	if utf8.ValidString(s) && strings.IndexFunc(s, rawInvalid) < 0 {
 		return "`" + s + "`"
 	}
 	return qs
@@ -256,23 +400,3 @@
 	}
 	return fmt.Sprintf(f, u)
 }
-
-// formatPointer prints the address of the pointer.
-func formatPointer(v reflect.Value) string {
-	p := v.Pointer()
-	if flags.Deterministic {
-		p = 0xdeadf00f // Only used for stable testing purposes
-	}
-	return fmt.Sprintf("⟪0x%x⟫", p)
-}
-
-type visitedPointers map[value.Pointer]struct{}
-
-// Visit inserts pointer v into the visited map and reports whether it had
-// already been visited before.
-func (m visitedPointers) Visit(v reflect.Value) bool {
-	p := value.PointerOf(v)
-	_, visited := m[p]
-	m[p] = struct{}{}
-	return visited
-}
diff --git a/vendor/github.com/google/go-cmp/cmp/report_slices.go b/vendor/github.com/google/go-cmp/cmp/report_slices.go
index eafcf2e..168f92f 100644
--- a/vendor/github.com/google/go-cmp/cmp/report_slices.go
+++ b/vendor/github.com/google/go-cmp/cmp/report_slices.go
@@ -1,6 +1,6 @@
 // Copyright 2019, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package cmp
 
@@ -8,6 +8,7 @@
 	"bytes"
 	"fmt"
 	"reflect"
+	"strconv"
 	"strings"
 	"unicode"
 	"unicode/utf8"
@@ -23,14 +24,35 @@
 		return false // Must be formatting in diff mode
 	case v.NumDiff == 0:
 		return false // No differences detected
-	case v.NumIgnored+v.NumCompared+v.NumTransformed > 0:
-		// TODO: Handle the case where someone uses bytes.Equal on a large slice.
-		return false // Some custom option was used to determined equality
 	case !v.ValueX.IsValid() || !v.ValueY.IsValid():
 		return false // Both values must be valid
+	case v.NumIgnored > 0:
+		return false // Some ignore option was used
+	case v.NumTransformed > 0:
+		return false // Some transform option was used
+	case v.NumCompared > 1:
+		return false // More than one comparison was used
+	case v.NumCompared == 1 && v.Type.Name() != "":
+		// The need for cmp to check applicability of options on every element
+		// in a slice is a significant performance detriment for large []byte.
+		// The workaround is to specify Comparer(bytes.Equal),
+		// which enables cmp to compare []byte more efficiently.
+		// If they differ, we still want to provide batched diffing.
+		// The logic disallows named types since they tend to have their own
+		// String method, with nicer formatting than what this provides.
+		return false
 	}
 
-	switch t := v.Type; t.Kind() {
+	// Check whether this is an interface with the same concrete types.
+	t := v.Type
+	vx, vy := v.ValueX, v.ValueY
+	if t.Kind() == reflect.Interface && !vx.IsNil() && !vy.IsNil() && vx.Elem().Type() == vy.Elem().Type() {
+		vx, vy = vx.Elem(), vy.Elem()
+		t = vx.Type()
+	}
+
+	// Check whether we provide specialized diffing for this type.
+	switch t.Kind() {
 	case reflect.String:
 	case reflect.Array, reflect.Slice:
 		// Only slices of primitive types have specialized handling.
@@ -42,6 +64,11 @@
 			return false
 		}
 
+		// Both slice values have to be non-empty.
+		if t.Kind() == reflect.Slice && (vx.Len() == 0 || vy.Len() == 0) {
+			return false
+		}
+
 		// If a sufficient number of elements already differ,
 		// use specialized formatting even if length requirement is not met.
 		if v.NumDiff > v.NumSame {
@@ -53,7 +80,7 @@
 
 	// Use specialized string diffing for longer slices or strings.
 	const minLength = 64
-	return v.ValueX.Len() >= minLength && v.ValueY.Len() >= minLength
+	return vx.Len() >= minLength && vy.Len() >= minLength
 }
 
 // FormatDiffSlice prints a diff for the slices (or strings) represented by v.
@@ -62,6 +89,11 @@
 func (opts formatOptions) FormatDiffSlice(v *valueNode) textNode {
 	assert(opts.DiffMode == diffUnknown)
 	t, vx, vy := v.Type, v.ValueX, v.ValueY
+	if t.Kind() == reflect.Interface {
+		vx, vy = vx.Elem(), vy.Elem()
+		t = vx.Type()
+		opts = opts.WithTypeMode(emitType)
+	}
 
 	// Auto-detect the type of the data.
 	var isLinedText, isText, isBinary bool
@@ -82,7 +114,7 @@
 	}
 	if isText || isBinary {
 		var numLines, lastLineIdx, maxLineLen int
-		isBinary = false
+		isBinary = !utf8.ValidString(sx) || !utf8.ValidString(sy)
 		for i, r := range sx + sy {
 			if !(unicode.IsPrint(r) || unicode.IsSpace(r)) || r == utf8.RuneError {
 				isBinary = true
@@ -97,7 +129,7 @@
 			}
 		}
 		isText = !isBinary
-		isLinedText = isText && numLines >= 4 && maxLineLen <= 256
+		isLinedText = isText && numLines >= 4 && maxLineLen <= 1024
 	}
 
 	// Format the string into printable records.
@@ -117,6 +149,83 @@
 			},
 		)
 		delim = "\n"
+
+		// If possible, use a custom triple-quote (""") syntax for printing
+		// differences in a string literal. This format is more readable,
+		// but has edge-cases where differences are visually indistinguishable.
+		// This format is avoided under the following conditions:
+		//	• A line starts with `"""`
+		//	• A line starts with "..."
+		//	• A line contains non-printable characters
+		//	• Adjacent different lines differ only by whitespace
+		//
+		// For example:
+		//		"""
+		//		... // 3 identical lines
+		//		foo
+		//		bar
+		//	-	baz
+		//	+	BAZ
+		//		"""
+		isTripleQuoted := true
+		prevRemoveLines := map[string]bool{}
+		prevInsertLines := map[string]bool{}
+		var list2 textList
+		list2 = append(list2, textRecord{Value: textLine(`"""`), ElideComma: true})
+		for _, r := range list {
+			if !r.Value.Equal(textEllipsis) {
+				line, _ := strconv.Unquote(string(r.Value.(textLine)))
+				line = strings.TrimPrefix(strings.TrimSuffix(line, "\r"), "\r") // trim leading/trailing carriage returns for legacy Windows endline support
+				normLine := strings.Map(func(r rune) rune {
+					if unicode.IsSpace(r) {
+						return -1 // drop whitespace to avoid visually indistinguishable output
+					}
+					return r
+				}, line)
+				isPrintable := func(r rune) bool {
+					return unicode.IsPrint(r) || r == '\t' // specially treat tab as printable
+				}
+				isTripleQuoted = !strings.HasPrefix(line, `"""`) && !strings.HasPrefix(line, "...") && strings.TrimFunc(line, isPrintable) == ""
+				switch r.Diff {
+				case diffRemoved:
+					isTripleQuoted = isTripleQuoted && !prevInsertLines[normLine]
+					prevRemoveLines[normLine] = true
+				case diffInserted:
+					isTripleQuoted = isTripleQuoted && !prevRemoveLines[normLine]
+					prevInsertLines[normLine] = true
+				}
+				if !isTripleQuoted {
+					break
+				}
+				r.Value = textLine(line)
+				r.ElideComma = true
+			}
+			if !(r.Diff == diffRemoved || r.Diff == diffInserted) { // start a new non-adjacent difference group
+				prevRemoveLines = map[string]bool{}
+				prevInsertLines = map[string]bool{}
+			}
+			list2 = append(list2, r)
+		}
+		if r := list2[len(list2)-1]; r.Diff == diffIdentical && len(r.Value.(textLine)) == 0 {
+			list2 = list2[:len(list2)-1] // elide single empty line at the end
+		}
+		list2 = append(list2, textRecord{Value: textLine(`"""`), ElideComma: true})
+		if isTripleQuoted {
+			var out textNode = &textWrap{Prefix: "(", Value: list2, Suffix: ")"}
+			switch t.Kind() {
+			case reflect.String:
+				if t != reflect.TypeOf(string("")) {
+					out = opts.FormatType(t, out)
+				}
+			case reflect.Slice:
+				// Always emit type for slices since the triple-quote syntax
+				// looks like a string (not a slice).
+				opts = opts.WithTypeMode(emitType)
+				out = opts.FormatType(t, out)
+			}
+			return out
+		}
+
 	// If the text appears to be single-lined text,
 	// then perform differencing in approximately fixed-sized chunks.
 	// The output is printed as quoted strings.
@@ -129,6 +238,7 @@
 			},
 		)
 		delim = ""
+
 	// If the text appears to be binary data,
 	// then perform differencing in approximately fixed-sized chunks.
 	// The output is inspired by hexdump.
@@ -145,6 +255,7 @@
 				return textRecord{Diff: d, Value: textLine(s), Comment: comment}
 			},
 		)
+
 	// For all other slices of primitive types,
 	// then perform differencing in approximately fixed-sized chunks.
 	// The size of each chunk depends on the width of the element kind.
@@ -172,7 +283,9 @@
 					switch t.Elem().Kind() {
 					case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
 						ss = append(ss, fmt.Sprint(v.Index(i).Int()))
-					case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64, reflect.Uintptr:
+					case reflect.Uint, reflect.Uint16, reflect.Uint32, reflect.Uint64:
+						ss = append(ss, fmt.Sprint(v.Index(i).Uint()))
+					case reflect.Uint8, reflect.Uintptr:
 						ss = append(ss, formatHex(v.Index(i).Uint()))
 					case reflect.Bool, reflect.Float32, reflect.Float64, reflect.Complex64, reflect.Complex128:
 						ss = append(ss, fmt.Sprint(v.Index(i).Interface()))
@@ -185,7 +298,7 @@
 	}
 
 	// Wrap the output with appropriate type information.
-	var out textNode = textWrap{"{", list, "}"}
+	var out textNode = &textWrap{Prefix: "{", Value: list, Suffix: "}"}
 	if !isText {
 		// The "{...}" byte-sequence literal is not valid Go syntax for strings.
 		// Emit the type for extra clarity (e.g. "string{...}").
@@ -196,12 +309,12 @@
 	}
 	switch t.Kind() {
 	case reflect.String:
-		out = textWrap{"strings.Join(", out, fmt.Sprintf(", %q)", delim)}
+		out = &textWrap{Prefix: "strings.Join(", Value: out, Suffix: fmt.Sprintf(", %q)", delim)}
 		if t != reflect.TypeOf(string("")) {
 			out = opts.FormatType(t, out)
 		}
 	case reflect.Slice:
-		out = textWrap{"bytes.Join(", out, fmt.Sprintf(", %q)", delim)}
+		out = &textWrap{Prefix: "bytes.Join(", Value: out, Suffix: fmt.Sprintf(", %q)", delim)}
 		if t != reflect.TypeOf([]byte(nil)) {
 			out = opts.FormatType(t, out)
 		}
@@ -242,9 +355,22 @@
 		return n0 - v.Len()
 	}
 
+	var numDiffs int
+	maxLen := -1
+	if opts.LimitVerbosity {
+		maxLen = (1 << opts.verbosity()) << 2 // 4, 8, 16, 32, 64, etc...
+		opts.VerbosityLevel--
+	}
+
 	groups := coalesceAdjacentEdits(name, es)
 	groups = coalesceInterveningIdentical(groups, chunkSize/4)
+	maxGroup := diffStats{Name: name}
 	for i, ds := range groups {
+		if maxLen >= 0 && numDiffs >= maxLen {
+			maxGroup = maxGroup.Append(ds)
+			continue
+		}
+
 		// Print equal.
 		if ds.NumDiff() == 0 {
 			// Compute the number of leading and trailing equal bytes to print.
@@ -273,12 +399,18 @@
 		}
 
 		// Print unequal.
+		len0 := len(list)
 		nx := appendChunks(vx.Slice(0, ds.NumIdentical+ds.NumRemoved+ds.NumModified), diffRemoved)
 		vx = vx.Slice(nx, vx.Len())
 		ny := appendChunks(vy.Slice(0, ds.NumIdentical+ds.NumInserted+ds.NumModified), diffInserted)
 		vy = vy.Slice(ny, vy.Len())
+		numDiffs += len(list) - len0
 	}
-	assert(vx.Len() == 0 && vy.Len() == 0)
+	if maxGroup.IsZero() {
+		assert(vx.Len() == 0 && vy.Len() == 0)
+	} else {
+		list.AppendEllipsis(maxGroup)
+	}
 	return list
 }
 
diff --git a/vendor/github.com/google/go-cmp/cmp/report_text.go b/vendor/github.com/google/go-cmp/cmp/report_text.go
index 8b8fcab..0fd46d7 100644
--- a/vendor/github.com/google/go-cmp/cmp/report_text.go
+++ b/vendor/github.com/google/go-cmp/cmp/report_text.go
@@ -1,6 +1,6 @@
 // Copyright 2019, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package cmp
 
@@ -10,12 +10,15 @@
 	"math/rand"
 	"strings"
 	"time"
+	"unicode/utf8"
 
 	"github.com/google/go-cmp/cmp/internal/flags"
 )
 
 var randBool = rand.New(rand.NewSource(time.Now().Unix())).Intn(2) == 0
 
+const maxColumnLength = 80
+
 type indentMode int
 
 func (n indentMode) appendIndent(b []byte, d diffMode) []byte {
@@ -91,21 +94,22 @@
 // textWrap is a wrapper that concatenates a prefix and/or a suffix
 // to the underlying node.
 type textWrap struct {
-	Prefix string   // e.g., "bytes.Buffer{"
-	Value  textNode // textWrap | textList | textLine
-	Suffix string   // e.g., "}"
+	Prefix   string      // e.g., "bytes.Buffer{"
+	Value    textNode    // textWrap | textList | textLine
+	Suffix   string      // e.g., "}"
+	Metadata interface{} // arbitrary metadata; has no effect on formatting
 }
 
-func (s textWrap) Len() int {
+func (s *textWrap) Len() int {
 	return len(s.Prefix) + s.Value.Len() + len(s.Suffix)
 }
-func (s1 textWrap) Equal(s2 textNode) bool {
-	if s2, ok := s2.(textWrap); ok {
+func (s1 *textWrap) Equal(s2 textNode) bool {
+	if s2, ok := s2.(*textWrap); ok {
 		return s1.Prefix == s2.Prefix && s1.Value.Equal(s2.Value) && s1.Suffix == s2.Suffix
 	}
 	return false
 }
-func (s textWrap) String() string {
+func (s *textWrap) String() string {
 	var d diffMode
 	var n indentMode
 	_, s2 := s.formatCompactTo(nil, d)
@@ -114,7 +118,7 @@
 	b = append(b, '\n')              // Trailing newline
 	return string(b)
 }
-func (s textWrap) formatCompactTo(b []byte, d diffMode) ([]byte, textNode) {
+func (s *textWrap) formatCompactTo(b []byte, d diffMode) ([]byte, textNode) {
 	n0 := len(b) // Original buffer length
 	b = append(b, s.Prefix...)
 	b, s.Value = s.Value.formatCompactTo(b, d)
@@ -124,7 +128,7 @@
 	}
 	return b, s
 }
-func (s textWrap) formatExpandedTo(b []byte, d diffMode, n indentMode) []byte {
+func (s *textWrap) formatExpandedTo(b []byte, d diffMode, n indentMode) []byte {
 	b = append(b, s.Prefix...)
 	b = s.Value.formatExpandedTo(b, d, n)
 	b = append(b, s.Suffix...)
@@ -136,22 +140,23 @@
 // of the textList.formatCompactTo method.
 type textList []textRecord
 type textRecord struct {
-	Diff    diffMode     // e.g., 0 or '-' or '+'
-	Key     string       // e.g., "MyField"
-	Value   textNode     // textWrap | textLine
-	Comment fmt.Stringer // e.g., "6 identical fields"
+	Diff       diffMode     // e.g., 0 or '-' or '+'
+	Key        string       // e.g., "MyField"
+	Value      textNode     // textWrap | textLine
+	ElideComma bool         // avoid trailing comma
+	Comment    fmt.Stringer // e.g., "6 identical fields"
 }
 
 // AppendEllipsis appends a new ellipsis node to the list if none already
 // exists at the end. If cs is non-zero it coalesces the statistics with the
 // previous diffStats.
 func (s *textList) AppendEllipsis(ds diffStats) {
-	hasStats := ds != diffStats{}
+	hasStats := !ds.IsZero()
 	if len(*s) == 0 || !(*s)[len(*s)-1].Value.Equal(textEllipsis) {
 		if hasStats {
-			*s = append(*s, textRecord{Value: textEllipsis, Comment: ds})
+			*s = append(*s, textRecord{Value: textEllipsis, ElideComma: true, Comment: ds})
 		} else {
-			*s = append(*s, textRecord{Value: textEllipsis})
+			*s = append(*s, textRecord{Value: textEllipsis, ElideComma: true})
 		}
 		return
 	}
@@ -191,7 +196,7 @@
 }
 
 func (s textList) String() string {
-	return textWrap{"{", s, "}"}.String()
+	return (&textWrap{Prefix: "{", Value: s, Suffix: "}"}).String()
 }
 
 func (s textList) formatCompactTo(b []byte, d diffMode) ([]byte, textNode) {
@@ -221,7 +226,7 @@
 	}
 	// Force multi-lined output when printing a removed/inserted node that
 	// is sufficiently long.
-	if (d == diffInserted || d == diffRemoved) && len(b[n0:]) > 80 {
+	if (d == diffInserted || d == diffRemoved) && len(b[n0:]) > maxColumnLength {
 		multiLine = true
 	}
 	if !multiLine {
@@ -236,16 +241,50 @@
 			_, isLine := r.Value.(textLine)
 			return r.Key == "" || !isLine
 		},
-		func(r textRecord) int { return len(r.Key) },
+		func(r textRecord) int { return utf8.RuneCountInString(r.Key) },
 	)
 	alignValueLens := s.alignLens(
 		func(r textRecord) bool {
 			_, isLine := r.Value.(textLine)
 			return !isLine || r.Value.Equal(textEllipsis) || r.Comment == nil
 		},
-		func(r textRecord) int { return len(r.Value.(textLine)) },
+		func(r textRecord) int { return utf8.RuneCount(r.Value.(textLine)) },
 	)
 
+	// Format lists of simple lists in a batched form.
+	// If the list is sequence of only textLine values,
+	// then batch multiple values on a single line.
+	var isSimple bool
+	for _, r := range s {
+		_, isLine := r.Value.(textLine)
+		isSimple = r.Diff == 0 && r.Key == "" && isLine && r.Comment == nil
+		if !isSimple {
+			break
+		}
+	}
+	if isSimple {
+		n++
+		var batch []byte
+		emitBatch := func() {
+			if len(batch) > 0 {
+				b = n.appendIndent(append(b, '\n'), d)
+				b = append(b, bytes.TrimRight(batch, " ")...)
+				batch = batch[:0]
+			}
+		}
+		for _, r := range s {
+			line := r.Value.(textLine)
+			if len(batch)+len(line)+len(", ") > maxColumnLength {
+				emitBatch()
+			}
+			batch = append(batch, line...)
+			batch = append(batch, ", "...)
+		}
+		emitBatch()
+		n--
+		return n.appendIndent(append(b, '\n'), d)
+	}
+
 	// Format the list as a multi-lined output.
 	n++
 	for i, r := range s {
@@ -256,7 +295,7 @@
 		b = alignKeyLens[i].appendChar(b, ' ')
 
 		b = r.Value.formatExpandedTo(b, d|r.Diff, n)
-		if !r.Value.Equal(textEllipsis) {
+		if !r.ElideComma {
 			b = append(b, ',')
 		}
 		b = alignValueLens[i].appendChar(b, ' ')
@@ -332,6 +371,11 @@
 	NumModified  int
 }
 
+func (s diffStats) IsZero() bool {
+	s.Name = ""
+	return s == diffStats{}
+}
+
 func (s diffStats) NumDiff() int {
 	return s.NumRemoved + s.NumInserted + s.NumModified
 }
diff --git a/vendor/github.com/google/go-cmp/cmp/report_value.go b/vendor/github.com/google/go-cmp/cmp/report_value.go
index 83031a7..668d470 100644
--- a/vendor/github.com/google/go-cmp/cmp/report_value.go
+++ b/vendor/github.com/google/go-cmp/cmp/report_value.go
@@ -1,6 +1,6 @@
 // Copyright 2019, The Go Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE.md file.
+// license that can be found in the LICENSE file.
 
 package cmp