gh-131196: Add C implementation of UUID stringification

[akx]

This PR follows up on #140663 by, well, following up on my musing about whether there should be an optimized C implementation of UUID stringification.

I tried to tighten the C implementation as much as possible, e.g. by directly writing into an 1BYTE_KIND PyUnicode object (as e.g. pystrhex.c does, which I couldn't use here because of the unique separator situation UUIDs have). I'd appreciate extra eyes on it, though!

The speedup seems pleasant for a fairly small patch – on my machine, some 4.5x:

$ ./python.exe -m timeit -s 'import uuid; uuid.UUID.__str__ = uuid._py_uuid_str_method; u = uuid.uuid4()' 'str(u)'
500000 loops, best of 5: 443 nsec per loop
$ ./python.exe -m timeit -s 'import uuid; u = uuid.uuid4()' 'str(u)'
2000000 loops, best of 5: 97.2 nsec per loop

• Issue: Improve perfomance of UUID.hex and UUID.__str__ by using bytes.hex() #131196

[akx]

Would be superseded by #139123 which is far further along in review, so closing :)