Add new vector-search-bench benchmarking crate

[connortsui20]

Summary

Tracking issue: #7297

Adds a new vector-search-bench crate.

Right now this is just utilities is preparation for adding proper vector search benchmarks (that are pulled from disk, not just in-memory). This just includes data downloading and file preparation, conversion (from parquet lists to vortex vector arrays), and some different compression and scan utilities.

Testing

Some basic unit tests but the real stress test will come later when we actually benchmark stuff.

[codspeed-hq]

Merging this PR will improve performance by 22.08%

⚡ 10 improved benchmarks
✅ 1153 untouched benchmarks
⏩ 1457 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	take_map[(0.1, 0.5)]	1,154.5 µs	990.4 µs	+16.57%
⚡	Simulation	take_map[(0.1, 1.0)]	2 ms	1.7 ms	+19.06%
⚡	Simulation	old_alp_prim_test_between[f64, 32768]	347.6 µs	284.7 µs	+22.08%
⚡	Simulation	patched_take_10k_contiguous_patches	258.1 µs	227.7 µs	+13.32%
⚡	Simulation	patched_take_10k_dispersed	316 µs	285.8 µs	+10.58%
⚡	Simulation	patched_take_10k_random	270.3 µs	240 µs	+12.64%
⚡	Simulation	patched_take_10k_contiguous_not_patches	258.4 µs	228.1 µs	+13.29%
⚡	Simulation	take_10k_dispersed	284.4 µs	239.5 µs	+18.76%
⚡	Simulation	patched_take_10k_first_chunk_only	302 µs	271.8 µs	+11.14%
⚡	Simulation	take_10k_first_chunk_only	270.6 µs	225.7 µs	+19.89%

---

_{Comparing ct/parquet-to-vortex-vector (3c957e3) with develop (8b09d0c)}

Footnotes

1. 1457 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩