Add missing FIPS 186-5 GCD check in Lucas probable prime test

[Shounak611]

Implements the additional FIPS 186-5 B.3.3 check:

gcd(C, (1-D)/4) == 1

inside is_lucas_probable_prime().

This aligns the implementation with the updated FIPS 186-5 specification.

[randombit]

I would at the very least want some testcases that actually hits this. (1-D)/4) is going to just be 1,2,3,... afaict this is basically just testing (in a somewhat expensive way) that Cdoesn't have any very small factors. And I'm pretty sure our existing check that jacobi(C,D) != 0 catches this anyway.

If needed, it would be much more effective to precompute 2*3*5*... up to some reasonable size then gcd C with that. That GCD would eliminate strictly more values of C than this check would.

[Shounak611]

I’ll verify whether the additional GCD check actually rejects any cases not already handled by the existing Jacobi condition and try to produce concrete examples/benchmarks. If it turns out to be redundant or not materially beneficial, I’ll remove the change from the PR.