test how an implementation handles MPI(0)
openpgp-wg/rfc4880bis#62 (closed) describes a crash in a popular OpenPGP implementation when encountering a zero-length MPI.
We should probably test how well an implementation handles such an object.
Most such objects will be invalid, which means i guess that the test would need to be able to distinguish between a crash and a legitimate failure to deal with the object.
Possibly some very unlucky legitimate RSA signature could end up with a value of 0, but crafting such an object would be pretty expensive.