gh-119182: Optimize PyUnicode_FromFormat()

[vstinner]

Use strchr() and ucs1lib_find_max_char() to optimize the code path formatting sub-strings between '%' formats.

• Issue: [C API] Add an efficient public PyUnicodeWriter API #119182

[vstinner]

Results of #120248 (comment) benchmark:

$ python3 -m pyperf compare_to ref.json change.json --table
+----------------+--------+----------------------+
| Benchmark      | ref    | change               |
+================+========+======================+
| bench 30       | 219 ns | 202 ns: 1.08x faster |
+----------------+--------+----------------------+
| bench 100      | 262 ns | 216 ns: 1.21x faster |
+----------------+--------+----------------------+
| Geometric mean | (ref)  | 1.09x faster         |
+----------------+--------+----------------------+

Benchmark hidden because not significant (1): bench 3

[serhiy-storchaka]

You can run it once, for the whole format string.

[vstinner]

Right. I modified my PR to run it only once, good idea!

[vstinner]

Merged, thanks for the reviews @serhiy-storchaka.