gh-117431: Improve performance of startswith, endswith, count and find methods for strings and bytes

[eendebakpt]

Benchmarks (non-PGO):

main: python -m timeit -s "s = 'abcdef'" "s.startswith('abc')"
5000000 loops, best of 5: 63.8 nsec per loop
pr: python -m timeit -s "s = 'abcdef'" "s.startswith('abc')"
5000000 loops, best of 5: 42.3 nsec per loop

main: python -m timeit -s "s = 'abcdef'" "s.count('a')"
5000000 loops, best of 5: 70 nsec per loop
pr: python -m timeit -s "s = 'abcdef'" "s.count('a')"
5000000 loops, best of 5: 51.4 nsec per loop

main: python -m timeit -s "s = b'abcdef'" "s.startswith(b'q')"
5000000 loops, best of 5: 68.3 nsec per loop
pr: python -m timeit -s "s = b'abcdef'" "s.startswith(b'q')"
5000000 loops, best of 5: 44.5 nsec per loop

Notes:

• We could also add a fast path for the case where there are 2 or 3 arguments. Since all arguments are positional only, for 1, 2 or 3 arguments the parsing cannot fail. That will increase performance for the 2 and 3 argument case, but it adds some more complexity.

• Issue: Improve performance of startswith, endswith, count, *find, and *index methods for str, bytes and bytearray #117431

[eendebakpt]

Closing in favor of #117466