gh-89381: Fix signature for math.log and cmath.log

[hauntsaninja]

The optional group in math.log seems unnecessary, as far as I can tell. Once we remove this, Argument Clinic knows what to do.

For cmath, we just do what math is already doing: use c_default = NULL (so the actual runtime logic is unchanged), but mark cmath.e as the fake default value.

• Issue: cmath.log has an invalid signature #89381

[skirpichev]

Seems to be working. I'll keep #101070 in case someone will prefer base=None defaults. Now we got
something like this:

>>> inspect.signature(math.log)
<Signature (x, base=2.718281828459045, /)>

Maybe this is slightly cryptic: 1) magic constant appear, 2) there is an illusion that the special base=math.e works internally as log(x)/log(base). BTW, shouldn't be there regression tests (e.g. for the original issue with inspect.signature(cmath.log))?

Also, I think that docs for the math.log/cmath.log should be adapted. To me it was a surprise, that this requires a manual adjustment after fixing the docstring.

[mdickinson]

Maybe this is slightly cryptic: 1) magic constant appear, 2) there is an illusion that the special base=math.e works internally as log(x)/log(base).

Agreed on both points; I prefer the solution in #101070.

[skirpichev]

@mdickinson, I can restore #101070, if you wish. But after #101123 I'm not sure that a broken signature is an issue for CPython.

[mdickinson]

@skirpichev The changes in #101070 look good to me, modulo one documentation suggestion (left in a comment). If you resurrect it, I'd be happy to merge it.