Use unaligned bit buffer iterators where possible

[robert3005]

Use unalingned bit iterators where possible, they are faster as they avoid
unaligned reads. They're always fine to use for unary operations and can be used
for n-ary operations if all buffers are aligned

Summary

We used to have this optimisation when we added BitBuffer but we reverted to
arrow implementation of bit iterators and it got lost in the process.

Closes: #6836

Testing

Existing tests

Signed-off-by: Robert Kruszewski github@robertk.io

[codspeed-hq]

Merging this PR will degrade performance by 20.89%

⚠️ 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

⚡ 14 improved benchmarks
❌ 5 regressed benchmarks
✅ 1010 untouched benchmarks
⏩ 1466 skipped benchmarks¹

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

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	chunked_varbinview_opt_canonical_into[(1000, 10)]	3.8 ms	3.3 ms	+15.82%
⚡	Simulation	chunked_varbinview_opt_canonical_into[(100, 100)]	4.2 ms	3.7 ms	+14.53%
⚡	Simulation	chunked_varbinview_opt_into_canonical[(100, 100)]	4.2 ms	3.7 ms	+14.41%
⚡	Simulation	chunked_varbinview_opt_into_canonical[(1000, 10)]	3.8 ms	3.3 ms	+15.77%
⚡	Simulation	varbinview_zip_block_mask	27.4 ms	24 ms	+14.29%
❌	Simulation	dict_canonicalize_zipfian[16, 1000]	56.2 µs	63.4 µs	-11.39%
⚡	Simulation	varbinview_zip_fragmented_mask	31.5 ms	28.1 ms	+12.18%
❌	Simulation	patched_take_200k_first_chunk_only	4.8 ms	5.4 ms	-10.68%
❌	Simulation	patched_take_200k_dispersed	4.7 ms	5.6 ms	-16.56%
❌	Simulation	take_200k_first_chunk_only	3.3 ms	4.2 ms	-20.89%
❌	Simulation	take_200k_dispersed	3.6 ms	4.5 ms	-19.6%
⚡	Simulation	bitwise_and_vortex_buffer[16384]	13.7 µs	11 µs	+23.77%
⚡	Simulation	bitwise_and_vortex_buffer[65536]	40.9 µs	29.7 µs	+37.42%
⚡	Simulation	bitwise_not_vortex_buffer[16384]	10.4 µs	9.3 µs	+12.14%
⚡	Simulation	bitwise_not_vortex_buffer[2048]	5.5 µs	4.6 µs	+18.18%
⚡	Simulation	bitwise_or_vortex_buffer[16384]	13.7 µs	11.1 µs	+23.71%
⚡	Simulation	bitwise_or_vortex_buffer[65536]	40.9 µs	29.8 µs	+37.38%
⚡	Simulation	bitwise_not_vortex_buffer[65536]	26.7 µs	21.5 µs	+24.23%
⚡	Simulation	from_iter_bit_buffer[128]	5.3 µs	4.8 µs	+10.18%

---

_{Comparing rk/faster_bit_buffer (1f588c9) with develop (11a2733)}

Footnotes

1. 1466 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. ↩

[robert3005]

I have to revert the last commit before merging