Should more equivalence interfaces be used in Assert.Equivalent?

[bradwilson]

Currently, Assert.Equivalent has a check for value types to see if they implement IComparable:

https://github.com/xunit/assert.xunit/blob/d04a4dcbb7288b4cf41e60db63a43be44f8d0cb5/Sdk/AssertHelper.cs#L124-L133

This was added to ensure that non-intrinsic value type equivalence wasn't broken when we merged a change to enable reference types embedded in value types: xunit/assert.xunit@4899f6e.

This logic could use some review. At least two questions come to mind:

1. Why does this only apply to value types and not reference types?

2. Why does this only check for IComparable? We current look for IEquatable<T>, IComparable<T>, and IComparable as well as IStructuralEquatable, IEquatable<T2>, and IComparable<T2> (there is an open issue about ordering)

This was discovered while working through issue #2696.