perf: optimize spark_size in spark-expr

[andygrove]

⚠️ Autonomously generated by an LLM. This PR was created end-to-end by an automated agent with no human author. It has passed automated correctness (unit tests + differential fuzz) and micro-benchmark gates, but requires full human review. Do not merge without expert scrutiny.

What changed

Optimizes spark_size in the native spark-expr crate.

Compute list row sizes from the offset buffer instead of allocating a sliced ArrayRef per row via list_array.value(i), removing one heap allocation per row.

Evidence

• Correctness: unit tests + seeded differential fuzz (bit-identical Arrow output vs main).
• Benchmark (criterion, Rust-only): see numbers below; gate required ≥ 5.0% mean improvement with non-overlapping confidence intervals and no regressions.

spark_size_ list of short arrays: 93.333% faster (base 151223ns -> cand 10082ns)
spark_size_ list of long arrays: 94.086% faster (base 156323ns -> cand 9245ns)

[mbutrovich]

LGTM! Thanks @andygrove!

[mbutrovich]

One question though, I can't find all of the benchmarks in the PR description in the code. Were they temporary, hallucinated, or am I missing something?

[andygrove]

One question though, I can't find all of the benchmarks in the PR description in the code. Were they temporary, hallucinated, or am I missing something?

Ah, this was from an earlier bug, now fixed, where benchmark results from different runs got mixed together. Let me re-verify this. Moving to draft.

[andygrove]

One question though, I can't find all of the benchmarks in the PR description in the code. Were they temporary, hallucinated, or am I missing something?

Ah, this was from an earlier bug, now fixed, where benchmark results from different runs got mixed together. Let me re-verify this. Moving to draft.

Updated. It was listing additional benchmarks from a different PR. I updated the PR description to remove those ones.