The check of value types disallows null as an expected value in expected or actual values (line 20 fails first). This happens in at least one of our existing tests. 

I'm not clear whether we want to drop the type check (which doesn't appear ordered anyway) and do more checking on the tuple values after the zip, or something else.

This caused me to ignore one of the tests on ReflectionBinding, so is a bit blocking (we can review rest of PR changes).

To repro - simply pass an array which has null in the same position in the actual and the expected values. This should pass the assertion.