feat: SQL compliance, robustness, and performance improvements

[subhayu99]

Summary

• SQL Compliance: Add HAVING clause, SELECT DISTINCT, FULL OUTER JOIN, IS NULL/IS NOT NULL support to the custom parser and executor. Fix NULL three-valued logic and empty GROUP BY aggregate defaults (e.g., SELECT COUNT(*) FROM empty.csv now returns 0).
• Robustness: Deduplicate _evaluate_condition from 5 copies across readers/operators into a single shared utility. Narrow broad except Exception handlers to specific types. Add column-not-found fuzzy matching with "Did you mean?" suggestions.
• Performance: Enable predicate pushdown for JOIN queries (left table columns only, conservative). Add condition reordering optimizer that evaluates equality conditions before range conditions. Sync version from importlib.metadata.

Test plan

• All 649 existing tests pass
• Ruff linter and formatter clean
• Manually test SELECT DISTINCT city FROM examples/employees.csv
• Manually test SELECT * FROM examples/employees.csv FULL OUTER JOIN examples/departments.csv ON department = name
• Manually test SELECT department, COUNT(*) FROM examples/employees.csv GROUP BY department HAVING COUNT(*) > 1
• Manually test SELECT * FROM examples/employees.csv WHERE nme = 'Alice' (should show fuzzy suggestion)

🤖 Generated with Claude Code