Pattern used for catching expected errors in tests should be able to verify for precise expected error

I've been using this pattern for negative tests, that are expecting a failure:

	defer func() { test.CatchExpectedError(recover(), t) }()

This pattern ends up causing a test to pass for any error, even an error that is an error in the test itself.  

Invent a pattern that ensures tests don't accidentally pass because the test itself is throwing an error.