Split bignum tests and re-do in python framework

[tom-cosgrove-arm]

A couple of significant changes need to be made to the bignum tests:

• The tests for bignum functions should be split into test cases by module (e.g. bignum_core, bignum_mod_raw, etc).
• Whenever possible, test cases should be generated automatically.

We have a PR for Python code to generate the test cases. Once it is merged, the split can be performed, and the test cases for at least the following bignum functions should be revisited:

• mbedtls_mpi_core_add_if()
• mbedtls_mpi_core_sub()
• mbedtls_mpi_core_mla()
• mbedtls_mpi_core_montmul()

[yanesca]

#6083 might be merged with some reservations regarding the test cases. The hope is that having automated test case generation in python will make them easier to review.

Either way, as part of this task, the test cases for #6083 need to be reviewed with full scrutiny, like they were newly added here.

Never mind, Gilles raised a separate issue for this (#6371).

[yanesca]

These tests were generated and the scripts and the approach can be found in these commit messages: 0cc7865 2a65b85 659c84a f334d96

The tests might have changed at later commits, but the generating principles and steps are a good starting point.