Fix iceberg column pruning

[sudsali]

Issue #, if available:

#667

Description of changes:

AnalysisRunner.runScanningAnalyzers runs .agg() on the full DataFrame without column pruning. V2 DataSource connectors (Iceberg, Delta Lake) make scan-planning decisions before Spark's optimizer simplifies the plan, causing full table reads on wide tables.

• Add columnsReferenced() method to Analyzer trait (default None : safe fallback)
• Override in all scanning analyzers to declare their column dependencies
• Add pruneColumns() in AnalysisRunner that selects only needed columns before .agg()
• Pruning is automatically disabled when any analyzer has a WHERE clause or free-form SQL predicate
• All 1018 tests pass including new tests for column pruning, where clause fallback, multi-column analyzers, and duplicate column handling

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.