Prune plan scans with zoned statistics

[joseph-isaacs]

Summary

• derive a statistics falsifier from each scan filter and optimize it as a physical plan
• bind the filter once and share that BoundExpression between exact filtering and pruning-plan
  construction
• add the static (ExpressionPlan, ZonedPlan) rule, representing pruning as bound state on
  ZonedPlan rather than introducing another plan-node type
• drop the zoned data child from pruning plans, cache the zone map, preserve unknown proof values,
  and expand zone proofs into the data row domain
• keep the temporary unbind only at the existing raw stats-rewrite adapter boundary
• execute pruning before constructing exact-filter or projection reads, so fully falsified splits
  do not request data
• preserve concurrent filter/projection registration so both consumers share an in-flight segment
  request
• cover before/after optimization, mixed row-index plus zoned struct-field partitioning, short
  final zones, skipped data-segment reads, and shared filter/projection reads

Stack

This is part 6 of the plan-native scan stack and is based on #9240. Its history is one coherent
pruning commit. The focused planning optimization is split into follow-up #9254.

Performance follow-up

The current implementation prioritizes correctness and keeps pruning inside the general array-plan
execution contract. On the 100-shard ClickBench scan, pruning eliminated 51.8% of rows and fully
skipped 264 of 1,100 splits, but prepared v2 execution was 44.41 ms with pruning versus 30.69 ms
without it. Profiling attributes roughly 5.3 ms to materializing a row-aligned boolean/validity
proof and another 8.4 ms to filtering the encoded input before the exact comparison.

A follow-up should make pruning mask-native and preserve encoded filter evaluation before combining
the pruning mask. That execution work is deliberately separate from this planning and correctness
PR.

Validation

• RUSTC_WRAPPER= cargo nextest run -p vortex-layout -p vortex-scan-v2
• RUSTC_WRAPPER= cargo clippy -p vortex-layout -p vortex-scan-v2 --all-targets --all-features -- -D warnings
• RUSTC_WRAPPER= cargo test --doc -p vortex-layout -p vortex-scan-v2
• cargo +nightly fmt --all -- --check
• git diff --check

[codspeed-hq]

Merging this PR will improve performance by 12.2%

⚡ 1 improved benchmark
✅ 1933 untouched benchmarks
⏩ 51 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	decompress[u64, (1000, 16)]	72.3 µs	64.4 µs	+12.2%

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 vortex-plan-zoned-pruning (1673868) with vortex-plan-scan-v2 (9260967)}

Footnotes

1. 51 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. ↩