gh-139156: Use PyBytesWriter in _PyUnicode_EncodeCharmap()

[vstinner]

Replace PyBytes_FromStringAndSize() and _PyBytes_Resize() with the PyBytesWriter API.

• Issue: Use PyBytesWriter in Unicode codecs #139156

[vstinner]

Benchmark:

import pyperf
runner = pyperf.Runner()
for size in (3, 100, 1000):
    runner.timeit(f'{size:,} ASCII chars',
        setup=f's="x"*{size}',
        stmt='s.encode("iso8859-15")')
for size in (3, 100, 1000):
    runner.timeit(f'{size:,} UCS-2 chars',
        setup=f's="€"*{size}',
        stmt='s.encode("iso8859-15")')

Results:

Benchmark	ref	pep782
3 ASCII chars	654 ns	674 ns: 1.03x slower
100 ASCII chars	1.11 us	1.13 us: 1.02x slower
1,000 ASCII chars	5.32 us	5.81 us: 1.09x slower
3 UCS-2 chars	659 ns	678 ns: 1.03x slower
100 UCS-2 chars	1.16 us	1.19 us: 1.02x slower
1,000 UCS-2 chars	5.81 us	6.23 us: 1.07x slower
Geometric mean	(ref)	1.04x slower

[vstinner]

Results on PGO+LTO build with CPU isolation:

Benchmark	ref	pep782
3 ASCII chars	563 ns	583 ns: 1.04x slower
100 ASCII chars	1.27 us	1.35 us: 1.06x slower
1,000 ASCII chars	7.87 us	8.77 us: 1.12x slower
3 UCS-2 chars	563 ns	585 ns: 1.04x slower
100 UCS-2 chars	1.33 us	1.38 us: 1.03x slower
1,000 UCS-2 chars	8.31 us	9.18 us: 1.11x slower
Geometric mean	(ref)	1.06x slower