gh-132284: Don't wrap base PyCFunction slots on class creation if not overridden

[tom-pytel]

Will not wrap slot functions from base classes if they are not explicitly overridden on class creations.

Subclass __getitem__() performance even increases a bit from:

$ ./python -m timeit -s $'class d(dict): pass\ni = d({1:2})' "i[1]"
10000000 loops, best of 5: 28.5 nsec per loop

to:

$ ./python -m timeit -s $'class d(dict): pass\ni = d({1:2})' "i[1]"
10000000 loops, best of 5: 21.6 nsec per loop

Rationale for the checks is as follows:

• generic == NULL because not sure of behavior with slots that loop several times like tp_getattr, tp_setattr and tp_richcompare.
• Py_IS_TYPE(descr, &PyMethodDescr_Type) because only if setting pre-existing builtin C method.
• *ptr == ((PyMethodDescrObject *)descr)->d_method->ml_meth checks both *ptr != NULL and stuff like:

class C:
    __new__ = type.__dict__['__subclasscheck__']

• Issue: Py_TYPE(lhs)->tp_as_mapping->mp_subscript is not the same as Dict_Type.tp_as_mapping->mp_subscript when it should be #132284

[brandtbucher]

Do we have a test that something like this continues to work correctly? If not, maybe we should add one:

>>> class D(dict):
...     pass
...     
>>> d = D({None: None})
>>> assert d[None] is None
>>> D.__getitem__ = lambda self, item: 42
>>> assert d[None] == 42

[iritkatriel]

This is probably worth an entry in "what's new" in case someone notices something change for some edge case.

[iritkatriel]

Also note merge conflict.

[tom-pytel]

This is probably worth an entry in "what's new" in case someone notices something change for some edge case.

Have a look at addition to whatsnew. Not sure if is right place so if you want something else then tell me what to put where or feel free to change as needed.