Preserve valid alignment when slicing buffer handles

[danking]

Authored by Codex.

Summary

• Add BufferHandle::slice_with_alignment, which reports the strongest alignment still guaranteed by a sliced view while enforcing a requested minimum alignment.
• Use the new API when slicing primitive payload buffers so over-aligned buffers can be sliced at element-aligned offsets without preserving accidental allocation over-alignment.
• Use the new API when extracting serialized buffers before the existing explicit ensure_aligned step.

Call-site assessment

• PrimitiveArray slicing needed the fix; compressor sampling slices primitive arrays and was preserving 256-byte allocation alignment through f32 element-aligned views.
• SerializedArray segment extraction now avoids strict source alignment before the explicit descriptor alignment check/copy.
• Bool host slicing already uses unaligned byte slicing; the device path already relies on DeviceBuffer implementations to report view alignment.

Testing

• cargo check -p vortex-array
• cargo test -p vortex-array host_slice_reports_strongest_supported_view_alignment
• cargo test -p vortex-array slice_over_aligned_f32_buffer_at_f32_aligned_offset
• cargo test -p vortex-array slice

[codspeed-hq]

Merging this PR will improve performance by 12.3%

⚠️ 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

⚡ 1 improved benchmark
✅ 2129 untouched benchmarks
⏩ 106 skipped benchmarks¹
🗄️ 4 archived benchmarks run²

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	WallTime	multiply_shapes_neon[(128, PerRowPerRow)]	2.2 µs	2 µs	+12.3%

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing codex/primitive-slice-offset-alignment (49df2f7) with develop (7feec60)}

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. ↩

[0ax1]

We have the notion of minimum and preferred alignment ( see buffer_mut.rs::with_capacity_aligned). The preferred alignment, our 256 bytes default, should only matter for allocation but never for correctness. There are no SIMD kernels that rely on and exploit the overalignment. Your PR patches the issue but the real issue is that we somewhere pass the preferred alignment as the required alignment. Saying slice_with_alignment should not be necessary.

[danking]

see instead: https://github.com/spiraldb/spiraldb/pull/6769/changes