We think it would be strange to see an expression like A ? B : C where
A is not one bit wide. It found a few minor things that we were able to
clean up, but nothing that was really a bug.
Since the ?: operator has the lowest precedence, expressions like A
& B ? C : D are parsed as (A & B) ? C : D, which might not be what is
intended. In some cases, an actual precedence problem might be revealed by
seeing that the size of the test expression isn't 1.
