More robust handling of invalid spec test cases

[wemeetagain]

Hmmm, when a test is invalid, we expect process() to throw an error. But it could be an issue that it is throwing an error that is different from what we are expecting, eg. we expect attester_slashing / invalid_att1_empty_indices to throw error because of empty indices and not because of invalid signature.

We need to provide a mechanism to debug (maybe a simple log is fine?). Here we're just naively returning and swallow that error.

Originally posted by @ensi321 in #66 (comment)