Cache min/max indices in patches

[palaska]

Cache min/max patch indices in Patches and short-circuit search_index when the query falls outside that range. Speeds up ALP/BitPacked scalar_at. This should help with slicing any patched array into a region that doesn't overlap the patch range

[codspeed-hq]

Merging this PR will degrade performance by 64.19%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 4 improved benchmarks
❌ 12 regressed benchmarks
✅ 1164 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	search_index_above_max	266 µs	499.5 µs	-46.75%
⚡	Simulation	search_index_above_max_chunked	1,869.9 µs	499.5 µs	×3.7
❌	Simulation	search_index_below_min	262 µs	496.9 µs	-47.27%
⚡	Simulation	search_index_below_min_chunked	1,550.4 µs	496.9 µs	×3.1
❌	Simulation	search_index_full_range_random	264 µs	737.3 µs	-64.19%
❌	Simulation	search_index_full_range_random_chunked	1.6 ms	2.1 ms	-22.02%
❌	Simulation	search_index_in_range	264.3 µs	737.2 µs	-64.14%
❌	Simulation	search_index_in_range_chunked	2.1 ms	2.5 ms	-17.84%
❌	Simulation	search_index_mixed_out_of_range	264 µs	498.2 µs	-47%
⚡	Simulation	search_index_mixed_out_of_range_chunked	1,568.5 µs	498.2 µs	×3.1
❌	Simulation	patched_take_10k_dispersed	286.5 µs	344.9 µs	-16.93%
❌	Simulation	patched_take_10k_first_chunk_only	273 µs	331.3 µs	-17.6%
❌	Simulation	patched_take_10k_adversarial	229.6 µs	288.1 µs	-20.32%
❌	Simulation	take_10k_dispersed	240.8 µs	298.5 µs	-19.33%
❌	Simulation	take_10k_first_chunk_only	226.9 µs	284.2 µs	-20.18%
⚡	Simulation	decompress_rd[f64, (100000, 0.0)]	980.2 µs	843.6 µs	+16.18%

---

_{Comparing bp/minmax-index (6efcd3e) with develop (d7c22ba)}

[joseph-isaacs]

Indices arrays have min/max stats these should be used instead.

We might not currently save or use this, but by these benchmarks we should.

If these benchmarks are cleaned up I am happy to merge those first.

[joseph-isaacs]

See: https://github.com/vortex-data/vortex/blob/e03bd1dcc63a34f260a16c05395146e26d1c9df3/docs/developer-guide/benchmarking.md

[robert3005]

Meta point, this is why you want to have special logic that's shared across all arrays, if we have Patched array then that could be one place where all of this logic could live without hacks. Alas we need Patched array to get over the line