Strengthen signature verification failure tests

[SWilson4]

For background, please see this discussion: #1919 (comment) and related issue #1995.

It may be desirable to modify or extend our signature tests to take into account "weaker" forgeries as well as specific security claims (EUF-CMA vs SUF-CMA).

[rtjk]

If we wanted to thoroughly check trivial forgeries for SUF-CMA, instead of incrementing a single random byte we could have the test flip each bit of a signed message, one at a time, and expect a verification error. I made something similar to investigate #1995, would this be a welcome addition or is it out of scope?

[baentsch]

would this be a welcome addition or is it out of scope?

any improvement on testing the quality of code in OQS is welcome!

[dstebila]

@rtjk Your proposal above would be nice to include. Is it something you'd be able to spend some cycles on?

[rtjk]

@rtjk Your proposal above would be nice to include. Is it something you'd be able to spend some cycles on?

Sure! I can adapt the test I wrote for CROSS. Would it make sense to keep this in test_sig.c, or would it be better to create a separate file like test_sig_bitflip.c?

[dstebila]

@rtjk Your proposal above would be nice to include. Is it something you'd be able to spend some cycles on?

Sure! I can adapt the test I wrote for CROSS. Would it make sense to keep this in test_sig.c, or would it be better to create a separate file like test_sig_bitflip.c?

Great!

I think it would be okay to do all of it in test_sig.c.