Fuzzer: Ignore our current bug with the type of br_if

[kripken]

We give br_if a too specific type: #3767

This is only noticeable with GC, and in rare cases where the type of br_if
is actually used - which realistically it never is, so really just fuzzer testcases.

[tlively]

Would it be better to just relax our validation rules? That way at least we could have tests demonstrating our implemented behavior.

[kripken]

The validation error comes from v8 in this case. We emit code that is valid if the br_if's type is more specific, then v8 (or any other compliant VM) sees a too-general type.

Getting us to emit the proper general type is not trivial (it's not present on the br_if - it's only known by the block we branch to). But also as mentioned in the linked issue, we probably want to keep the specific type (as with tee)....

[tlively]

Oh, I see, we would have to insert extra downcasts to make V8 happy, but that's not worth it considering that V8 will probably be updated to accept what we are emitting. Makes sense 👍