perf(rust): reduce sort32 bucket overhead

[dexhunter]

Summary

• count valid keys in fixed-size chunks to reduce loop overhead
• reuse one mutable bucket reference per scatter instead of indexing each bucket twice
• return before the scatter passes when there are no active splats
• add regression tests for invalid keys, finite-value boundaries, and stable ordering

Performance

The benchmark builds the production Wasm sorter from a pinned upstream archive, overlays only rust/spark-rs/src/sort.rs, and compares baseline and candidate calls on the same process and fixtures. It validates exact active counts, descending order, stable equal-key ordering, unchanged input/padding, and Wasm size before printing the metric.

sort32_paired_latency_score is the candidate/baseline latency ratio, so lower is better and 1.0 is the baseline. Three final-patch confirmations measured 0.750563, 0.768874, and 0.760215. The median is 0.760215, a 24.0% latency reduction. The suite includes small inputs and one-million-key random, periodic, and adversarial layouts.

Autoresearch trajectory: https://dashboard.weco.ai/share/4avZ2RktRLkZmJ1sWBU1Tiqqd20Zndg7

Validation

• cargo test --manifest-path rust/spark-rs/Cargo.toml (2 passed)
• strict paired Wasm evaluator across 10 input layouts
• git diff --check