fix(buffer): don't create &mut [u64] over uninitialized memory in collect_words_in

[jianhe25]

BitBufferMut::collect_words_in allocated with with_capacity_in, called set_len to claim the memory was initialised, and passed as_mut_slice() to the fill closure. That creates a &mut [u64] over uninitialised memory before anything writes to it. The safety comment argued this was fine because fill writes every word and u64 has no invalid bit patterns, but a reference to uninitialised memory is UB on its own.

This uses BufferMut::zeroed_in(num_words, allocator) instead, keeping the allocator. fill still overwrites every word, so the only extra cost is one memset.

Testing

• Behaviour is unchanged and already covered by the existing BitBufferMut tests; I didn't add a new one.
• cargo test -p vortex-buffer --all-features, clippy, nightly fmt, and Miri on bit::buf_mut with CI's MIRIFLAGS.

AI assistance

This change was prepared with help from an AI coding assistant (Google-internal agentic tooling). I reviewed and verified the change and the tests myself.

[codspeed]

Merging this PR will degrade performance by 17.24%

⚠️ 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

❌ 2 regressed benchmarks
✅ 2101 untouched benchmarks
⏩ 503 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	from_bool_slice[65536]	74.6 µs	97.4 µs	-23.46%
❌	WallTime	dict_canonicalize_gt_u8_avx512[1000000]	422.3 µs	472 µs	-10.52%

Tip

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

---

_{Comparing jianhe25:buffer-fix-5 (0f4e9c6) with develop (b127616)}

Footnotes

1. 503 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. ↩