fix: support 64K alignment

[a10y]

Rationale for this change

Our Alignment type currently has a restriction that it fits in a u16. This is easy enough to satisfy most of the time.

However, while doing some testing with GH200 machines, which are ARM and which NVIDIA recommends you setup with 64K page alignment, I have seen some production code fail because it uses a custom allocator that passes back page-aligned memory, and attempting to construct 64K alignment is one past the maximum value so it panics.

What changes are included in this PR?

• Remove panic if Alignment > u16::MAX, now you can represent 64K alignment on buffers

What APIs are changed? Are there any user-facing changes?

Alignment now has no restriction that it must fit into a u16, so some code which may have previously panicked will no longer panic.

The infallable From<Alignment> for u16 impl has been removed entirely, there were zero callers within Vortex. This is technically a breaking change and anyone who was relying on this behavior should switch to using the exponent instead which will fit in u8.

[robert3005]

This is one chonky alignment

[codspeed-hq]

Merging this PR will not alter performance

⚠️ 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.

⚡ 7 improved benchmarks
❌ 6 regressed benchmarks
✅ 1576 untouched benchmarks
⏩ 4 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	copy_nullable[65536]	1 ms	1.4 ms	-24.26%
❌	Simulation	copy_non_nullable[65536]	908.6 µs	1,089.3 µs	-16.58%
❌	Simulation	slice_empty_vortex	310 ns	368.6 ns	-15.9%
❌	Simulation	chunked_varbinview_canonical_into[(100, 100)]	224.4 µs	259.5 µs	-13.55%
❌	Simulation	map_each[BufferMut<i32>, 128]	407.8 ns	466.1 ns	-12.51%
❌	Simulation	chunked_varbinview_into_canonical[(100, 100)]	271.5 µs	306.4 µs	-11.38%
⚡	Simulation	chunked_bool_canonical_into[(1000, 10)]	26.8 µs	16.1 µs	+66.54%
⚡	Simulation	bitwise_not_vortex_buffer_mut[128]	244.4 ns	186.1 ns	+31.34%
⚡	Simulation	bitwise_not_vortex_buffer_mut[1024]	304.7 ns	246.4 ns	+23.68%
⚡	Simulation	chunked_varbinview_into_canonical[(1000, 10)]	205.8 µs	168.9 µs	+21.85%
⚡	Simulation	bitwise_not_vortex_buffer_mut[2048]	398.6 ns	340.3 ns	+17.14%
⚡	Simulation	true_count_arrow_buffer[128]	784.7 ns	697.2 ns	+12.55%
⚡	Simulation	rebuild_naive	109.5 µs	98.8 µs	+10.79%

Tip

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

---

_{Comparing aduffy/allow-64k (5bec64a) with develop (3c53111)}

Footnotes

1. 4 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. ↩

[a10y]

take it up with Jensen