Add BoolArray::trim_bits, wire it to Canonical::compact that prunes unused bytes in the underlying buffer

[robert3005]

This is a cheap operation but not as cheap as to run it on construction of BoolArray.

Alternative to #8908

[codspeed-hq]

Merging this PR will degrade performance by 9.36%

⚠️ 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

⚡ 2 improved benchmarks
❌ 9 regressed benchmarks
✅ 1951 untouched benchmarks
⏩ 89 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	decompress[u64, (10000, 4)]	311.4 µs	402.3 µs	-22.59%
❌	Simulation	chunked_canonicalize_into[(1000, 50, 8, 4)]	12.9 ms	15.1 ms	-14.08%
❌	Simulation	chunked_canonicalize_into[(1000, 50, 8, 64)]	15.3 ms	17.7 ms	-13.44%
❌	Simulation	chunked_into_canonical[(1000, 50, 8, 64)]	17.1 ms	19.5 ms	-12.58%
❌	Simulation	search_index_in_range_chunked	5.3 ms	6.1 ms	-12.47%
❌	Simulation	chunked_canonicalize_into[(1000, 50, 8, 16)]	14 ms	16 ms	-12.07%
❌	Simulation	chunked_into_canonical[(1000, 50, 8, 4)]	14.5 ms	16.4 ms	-11.54%
❌	Simulation	chunked_into_canonical[(1000, 50, 8, 16)]	15.6 ms	17.5 ms	-10.74%
❌	Simulation	decompress[u64, (1000, 16)]	65.1 µs	72.8 µs	-10.59%
⚡	Simulation	take[core/shuffled/primitive/nonnull/chunks=1024/indices=10000]	14.6 ms	13.1 ms	+11.61%
⚡	Simulation	compress_fsst[(10000, 64, 8)]	10.8 ms	9.7 ms	+11.12%

Tip

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

---

_{Comparing rk/boolcompact (0baea1a) with develop (2d5b7e5)²}

Footnotes

1. 89 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. No successful run was found on develop (a0a0b20) during the generation of this report, so 2d5b7e5 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩