tracking: aggregate-stats scan optimization

[discord9]

Summary

Support aggregate-stats based scan optimization — when COUNT/MIN/MAX aggregates can be answered from parquet row-group/file statistics, skip normal row scans for eligible files and fall back safely when stats are missing or unsupported. This reduces scan cost for simple aggregate queries while preserving correctness through schema/type checks and row-scan fallback.

Related PRs

• feat: add aggregate-stats foundation types and helpers #8112 — feat: add aggregate-stats foundation types and helpers (reviewing)
• scanner-side stats runtime integration (not yet submitted)
• optimizer aggregate-stats rewrite (not yet submitted)
• sqlness and integration coverage (not yet submitted)

Component Breakdown

Component	Description	Status
Foundation types / helpers	Adds store-api stats DTOs, common-query stats candidate evaluation, and table row-group pruning stats helpers	🔄
Scanner runtime integration	Adds RegionScanner::scan_stats, stats-aware scanner properties/request wiring, runtime decision state, and mito2 stats stream production	🔜
Optimizer rewrite	Adds AggrStats optimizer rule and StatsScanExec to rewrite eligible aggregate scans to stats-backed execution	🔜
Validation coverage	Adds sqlness, integration tests, and partition/filter test adaptations for end-to-end behavior	🔜