[headtr1ck]

Use the vectorized functions of np.strings in the string accessor, now that the minimum numpy version is 2.2 (as suggested in #11722 (review)).

• len, startswith/endswith, find/rfind/index/rindex, strip, pad/center/ljust/rjust/zfill, repeat, the is* methods, and contains/replace with regex=False use np.strings for U, S and StringDType arrays; object arrays keep the elementwise path.

• 5-80x faster for 100k strings (new StringAccessor asv benchmark), unchanged for object arrays.

• Fixes results of e.g. strip, center and replace on dask arrays, which were truncated to a single character.

• Methods that are not C ufuncs in numpy 2.2 (upper, lower, translate, slice, ...) or use regular expressions stay elementwise.

• index on bytes now raises numpy's "substring not found" instead of "subsection not found".

• Tests added

• User visible changes (including notable bug fixes) are documented in whats-new.rst

AI Disclosure

• This PR contains AI-generated content.
  • I have tested any AI-generated content in my PR.
  • I take responsibility for any AI-generated content in my PR.
  • Tools: Claude Code with Opus 5.5

🤖 Generated with Claude Code

[github-actions]

🏎️ New personal best! Some benchmarks got better.

Change	Before [c52a59f]	After [b8c6750]	Ratio	Benchmark (Parameter)
🚀	28.1±0.2ms	1.54±0.02ms	0.05	accessors.StringAccessor.time_center('S')
🚀	33.3±0.1ms	1.38±0.05ms	0.04	accessors.StringAccessor.time_center('U')
🚀	32.7±0.3ms	2.67±0.01ms	0.08	accessors.StringAccessor.time_contains('S')
🚀	16.4±0.04ms	2.86±0.01ms	0.17	accessors.StringAccessor.time_contains('U')
🚀	27.0±0.3ms	2.63±0.01ms	0.1	accessors.StringAccessor.time_find('S')
🚀	27.5±0.2ms	2.84±0.01ms	0.1	accessors.StringAccessor.time_find('U')
🚀	11.7±0.09ms	2.02±0.01ms	0.17	accessors.StringAccessor.time_isalpha('S')
🚀	16.7±0.08ms	3.59±0.01ms	0.21	accessors.StringAccessor.time_isalpha('U')
🚀	7.55±0.04ms	212±4μs	0.03	accessors.StringAccessor.time_len('S')
🚀	11.2±0.2ms	230±20μs	0.02	accessors.StringAccessor.time_len('U')
🚀	21.3±0.1ms	1.20±0.01ms	0.06	accessors.StringAccessor.time_repeat('S')
🚀	27.2±0.09ms	1.41±0.05ms	0.05	accessors.StringAccessor.time_repeat('U')
🚀	26.2±0.5ms	4.09±0.05ms	0.16	accessors.StringAccessor.time_replace('S')
🚀	27.1±1ms	4.56±0.05ms	0.17	accessors.StringAccessor.time_replace('U')
🚀	17.5±0.5ms	551±7μs	0.03	accessors.StringAccessor.time_startswith('S')
🚀	17.9±0.09ms	571±8μs	0.03	accessors.StringAccessor.time_startswith('U')
🚀	18.5±0.1ms	1.67±0.02ms	0.09	accessors.StringAccessor.time_strip('S')
🚀	23.5±0.3ms	1.82±0.04ms	0.08	accessors.StringAccessor.time_strip('U')

🚀 better · 🐌 worse · 💥 now failing · 🩹 fixed · 🤷 not comparable

Comparing b8c6750d (this PR merged into the base branch) with c52a59f9 (base branch).
Only changes by more than a factor of 1.5 are shown, see the workflow run for all results.