perf: Optimize initcap()

[neilconway]

Which issue does this PR close?

• Closes Optimize initcap() #20351.

Rationale for this change

When all values in a Utf8/LargeUtf8 array are ASCII, we can skip using GenericStringBuilder and instead process the entire input buffer in a single pass using byte-level operations. This also avoids recomputing the offsets and nulls arrays. A similar optimization is already used for lower() and upper().

Along the way, optimize initcap_string() for ASCII-only inputs. It already had an ASCII-only fastpath but there was room for further optimization, by iterating over bytes rather than characters.

What changes are included in this PR?

• Cleanup benchmarks: we ran the scalar benchmark for different array sizes, despite the fact that it is invariant to the array size
• Add benchmark for different string lengths
• Add benchmark for Unicode array input
• Optimize for ASCII-only inputs as described above
• Add test case for ASCII-only input that is a sliced array

Are these changes tested?

Yes, plus an additional test added.

Are there any user-facing changes?

No.

[neilconway]

Benchmarks:

$ critcmp initcap-v2-vanilla initcap-v2-opt
group                                           initcap-v2-opt                         initcap-v2-vanilla
-----                                           --------------                         ------------------
initcap scalar/scalar_utf8                      1.00     99.3±0.82ns        ? ?/sec    1.53    151.7±0.87ns        ? ?/sec
initcap scalar/scalar_utf8view                  1.00     93.2±2.04ns        ? ?/sec    1.74    162.3±2.52ns        ? ?/sec
initcap size=1024 str_len=128/array_utf8        1.00     96.6±0.16µs        ? ?/sec    1.42    137.3±0.69µs        ? ?/sec
initcap size=1024 str_len=128/array_utf8view    1.00     73.2±0.05µs        ? ?/sec    1.81    132.3±0.24µs        ? ?/sec
initcap size=1024 str_len=16/array_utf8         1.00     13.6±0.02µs        ? ?/sec    1.55     21.0±0.06µs        ? ?/sec
initcap size=1024 str_len=16/array_utf8view     1.00     16.9±0.05µs        ? ?/sec    1.32     22.3±0.05µs        ? ?/sec
initcap size=4096 str_len=128/array_utf8        1.00    384.7±0.50µs        ? ?/sec    1.42    547.1±3.02µs        ? ?/sec
initcap size=4096 str_len=128/array_utf8view    1.00    291.6±0.28µs        ? ?/sec    1.82    529.4±3.35µs        ? ?/sec
initcap size=4096 str_len=16/array_utf8         1.00     52.9±0.07µs        ? ?/sec    1.56     82.5±0.14µs        ? ?/sec
initcap size=4096 str_len=16/array_utf8view     1.00     65.7±0.12µs        ? ?/sec    1.36     89.4±0.48µs        ? ?/sec
initcap size=8192 str_len=128/array_utf8        1.00    770.2±1.22µs        ? ?/sec    1.42   1096.4±2.68µs        ? ?/sec
initcap size=8192 str_len=128/array_utf8view    1.00    580.4±0.60µs        ? ?/sec    1.82   1059.1±5.74µs        ? ?/sec
initcap size=8192 str_len=16/array_utf8         1.00    105.4±0.38µs        ? ?/sec    1.58    166.3±0.41µs        ? ?/sec
initcap size=8192 str_len=16/array_utf8view     1.00    130.2±0.54µs        ? ?/sec    1.38    179.5±1.16µs        ? ?/sec