ALPRD: assert patch dtype instead of casting

[robert3005]

Stacked on #8610 (which is stacked on #8609).

ALPRD patch values are always stored as the non-nullable left-parts dtype, so the runtime cast in canonicalize_patches — which needed an ExecutionCtx to materialise a lazy cast plus an all_valid check — was unnecessary. This resolves the existing TODO(ngates)/TODO(joe): assert the DType, don't cast it.

Changes

• ALPRDData::canonicalize_patches(left_parts, patches, ctx) → ctx-free validate_patches(left_parts, patches) that asserts the patch values already have the non-nullable left-parts dtype (a non-nullable dtype also guarantees no nulls, so the all_valid execution is dropped).
• validate_parts now asserts the exact non-nullable patch dtype instead of eq_ignore_nullability + all_valid(LEGACY_SESSION).
• deserialize no longer mints a LEGACY_SESSION ctx for patch handling.
• take: Patches::take ignores null indices but reports a nullable dtype, so a lazy cast_values drops the spurious nullability (no execution at construction; materialised later during decode).

Net: removes 3 LEGACY_SESSION uses from ALPRD construction/validation, with no execution during construction.

ALP needed no change: its patch values are the original floats and its validate_patches already asserts the dtype (no cast). ALP::try_new / ALPRD::try_new public signatures are unchanged, so there is no caller cascade (CUDA/bench call sites are untouched).

Verification

• cargo build/clippy/test -p vortex-alp ✅ (173 tests, incl. take_with_nulls and take/decode conformance); clippy -D warnings clean; cargo +nightly fmt --check clean.
• cargo build --all-targets -p vortex-compressor -p vortex-btrblocks ✅ (ALP consumers).

🤖 Generated with Claude Code

https://claude.ai/code/session_01Mkrgj5SuJpaKBXK9jydtSc

---

Generated by Claude Code

[codspeed-hq]

Merging this PR will improve performance by 25.3%

⚠️ 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

⚡ 4 improved benchmarks
✅ 1585 untouched benchmarks
⏩ 10 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	chunked_bool_canonical_into[(1000, 10)]	27 µs	15.8 µs	+70.64%
⚡	Simulation	bitwise_not_vortex_buffer_mut[128]	215.3 ns	186.1 ns	+15.67%
⚡	Simulation	bitwise_not_vortex_buffer_mut[1024]	275.6 ns	246.4 ns	+11.84%
⚡	Simulation	compact_sliced[(4096, 90)]	838.1 ns	750.6 ns	+11.66%

Tip

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

---

_{Comparing claude/nice-hypatia-g68zgt-alp-patches (174c599) with claude/nice-hypatia-g68zgt-validate-ctx (7635ae6)}

Footnotes

1. 10 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. ↩