Add chunk-wise BitPackedV2 compression with PatchesV2

[joseph-isaacs]

Summary

• add FastLanes BitPackedV2 with a bit width per 1,024-value chunk
• add PatchesV2 with chunk-local indices and compact slot metadata
• use BitPackedV2 for unstable integer bitpacking and FoR residuals
• compress the three patch children adaptively instead of leaving them uncompressed

This is the base of a two-PR stack. It intentionally excludes BlockedFoR, FloatQuantV2, and temporary benchmark/scatter experiments.

Validation

• cargo +nightly fmt --all -- --check
• cargo check -p vortex-btrblocks
• cargo check -p vortex-btrblocks --features unstable_encodings
• cargo test -p vortex-btrblocks --features unstable_encodings test_bitpacking_v2_compresses_attached_patches
• stacked-tree targeted clippy: cargo clippy -p vortex-btrblocks -p vortex-bench --all-targets --features unstable_encodings -- -D warnings

Compression note

The prior combined branch reduced controlled TPC-H SF10 output from 2,455,096,884 to 2,416,147,124 bytes (-38,949,760, -1.5865%). That number is stack-level evidence, not an isolated claim for this PR; a clean branch-by-branch recompression should be used for final attribution.

[codspeed-hq]

Merging this PR will regress 2 benchmarks

⚠️ 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.

⚡ 4 improved benchmarks
❌ 2 regressed benchmarks
✅ 2151 untouched benchmarks
⏩ 106 skipped benchmarks¹
🗄️ 4 archived benchmarks run²

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	mul_i32_nonnull_avx512	7.2 µs	8.1 µs	-10.29%
❌	Simulation	compress_fsst[(500, 64, 8)]	482.6 µs	537.8 µs	-10.27%
⚡	WallTime	arrow_checked_add_u32_neon[16384]	20.5 µs	13.5 µs	+51.29%
⚡	WallTime	arrow_checked_add_u32_avx2[16384]	21.3 µs	17.7 µs	+20.37%
⚡	Simulation	take[duplicates/repeated/primitive/nonnull/chunks=16/indices=1000]	234.2 µs	201.2 µs	+16.37%
⚡	WallTime	words_gather_scalar_avx2[65536]	9.4 µs	8.3 µs	+13.63%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing ji/bitpacked-v2-patches (f4d3659) with develop (68e2aee)}

Footnotes

1. 106 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. ↩

2. 4 benchmarks were run, but are now archived. If they were deleted in another branch, consider rebasing to remove them from the report. Instead if they were added back, click here to restore them. ↩