feat(spark-3.5): [stacked] Add ANALYZE TABLE ... COMPUTE CLUSTERING QUALITY SQL extension

[mkuchenbecker]

OpenHouse spark-3.5 SQL Extensions Stack

PR	Base	Content
#660	main	Standalone refactor (decouple spark-3.5 from spark-3.1)
#661	#660	VACUUM
#662	#660	OPTIMIZE (bin-pack + incremental clustering)
(this)	#662	ANALYZE … COMPUTE CLUSTERING QUALITY

Depends on #662 (which depends on #660). This reuses the optimize.cluster.* property
contract and helpers introduced by OPTIMIZE, so it stacks on #662 rather than the
standalone refactor. Review/merge #660 → #662 → this.

Summary

Adds a read-only clustering-quality probe to the OpenHouse spark-3.5 SQL extensions:

ANALYZE TABLE <table> COMPUTE CLUSTERING QUALITY

Reports how well a table is clustered to its current key selection, using only what
OPTIMIZE persists (optimize.cluster.state) plus manifest metrics (t.files). No commit
and no property write. Emits (metric, dimension, value) rows: clustering_configured;
config_id / keys / sort_mode; coverage_bytes_pct / coverage_files_pct; per-key
depth_avg / depth_p90 / depth_max and the _covered variants (Snowflake-style
stabbing depth, the SLA quality input); null_bound_bytes_pct; unclustered_tail_hours;
state. Coverage is one aggregate over metadata and depth is a windowed sweep — both
distributed SQL, so the command is safe on tables with very large file counts.

Implemented as AnalyzeClusteringQualityExec, reusing the optimize.cluster.* contract
from OPTIMIZE, with an OpenHouse-table guard. Routing intercepts only the
COMPUTE CLUSTERING QUALITY variant, so ordinary ANALYZE TABLE … COMPUTE STATISTICS
still delegates to Spark.

Changes

• Client-facing API Changes
• Internal API Changes
• Bug Fixes
• New Features
• Performance Improvements
• Code Style
• Refactoring
• Documentation
• Tests

New non-reserved keywords ANALYZE / COMPUTE / CLUSTERING / QUALITY.

Testing Done

• Added new tests for the changes made.

• AnalyzeClusteringQualityExecTest (metricExpr / coveragePredicate) — 5 tests.

• AnalyzeClusteringQualityStatementTest (real Iceberg, Hadoop catalog): unconfigured flag,
  post-OPTIMIZE coverage/depth, read-only invariant, COMPUTE STATISTICS delegation,
  non-OpenHouse rejection — 5 tests.

./gradlew :integrations:spark:spark-3.5:openhouse-spark-3.5-runtime_2.12:test            # incl. 5 ANALYZE unit tests
./gradlew :integrations:spark:spark-3.5:openhouse-spark-3.5-itest:statementTest --tests '*AnalyzeClusteringQualityStatementTest'   # 5/5
# both BUILD SUCCESSFUL

Additional Information

• Large PR broken into smaller PRs, and PR plan linked in the description.