Adds tests against the wycheproof test vectors

[andozw]

Along the same spirit as the secp256k1 PR 1245. This tests against all 463 vectors but includes exclusions for 3 of them. Comments include the details outlining why each was excluded.

The json and c code was generated using the included python code (utils/tests_wycheproof_generate.py).

We decided not to duplicate this generation code for now.

[oreparaz]

LGTM. The 3 test cases we skip sound reasonable to me.

Can you please put a comment here with:

• old binary size
• new binary size
• difference in kB

Thanks!

[oreparaz]

LGTM. The 3 test cases we skip sound reasonable to me.

Can you please put a comment here with:

• old binary size
• new binary size
• difference in kB

Thanks!

[ivmaykov]

It looks like the subzero core binary failed to start in the background in the SignTx regression test action. Hard to tell what's going on, we probably need to improve the logic in the action so it's easier to get logs output. Did you run it locally with ASAN + UBSAN without any issues?

[andozw]

It looks like the subzero core binary failed to start in the background in the SignTx regression test action. Hard to tell what's going on, we probably need to improve the logic in the action so it's easier to get logs output. Did you run it locally with ASAN + UBSAN without any issues?

Ran locally with time(1). Took about 0.5 seconds without ASAN/UBSAN flags and about 2.6 seconds with ASAN/UBSAN.

Added sleeps in the CI script and the tests execute successfully which confirms the theory.

[andozw]

• old binary size
• new binary size
• difference in kB

Old binary size: 546916 bytes
New binary size: 629876 bytes
Difference ~82.9 kB

[ivmaykov]

Approving with some nits, please fix them before merging.

[ivmaykov]

Actually approving this time.

[ivmaykov]

LGTM