[PyRSA]

Summary

Add record-level expiration support for Primary Key table compaction, allowing expired records to be filtered during compaction and fully expired files to be removed through max-level rewrite.

Changes

• Add RecordLevelExpire to validate expiration configuration and filter expired records during compaction.
• Support INT, BIGINT, TIMESTAMP, and TIMESTAMP_LTZ expiration fields.
• Apply consistent expiration filtering to regular merge and Lookup compaction paths, avoiding expiration-only DELETE events.
• Select fully expired max-level files for expire-only rewrite when regular compaction has no pending tasks.
• Allow empty compaction output so fully expired input files can be retired.
• Add unit and integration tests covering expiration boundaries, Lookup behavior, file selection, and fully expired max-level rewrite.

Compatibility

• Expiration is applied during compaction; ordinary query results are not filtered.
• Append tables and Data Evolution compaction continue to reject record-level expiration.
• Files with historical Schema IDs remain conservatively ineligible for expire-only rewrite when their Stats cannot be resolved. Schema Evolution Stats compatibility is deferred to a follow-up PR.
• No expiration-only DELETE events are generated by Lookup compaction.

Testing

• Added expire_only_max_level_rewrite_retires_fully_expired_files to verify zero data-file output, input-file retirement, and no residual expired records.