feat(cascade): Extract cascade metadata to dedicated table

[arockwell]

Summary

• Moves cascade-specific fields (stage, pr_url) from documents table to dedicated document_cascade_metadata table
• Improves schema clarity and allows cascade features to evolve independently
• Maintains backward compatibility with dual-write strategy

Changes

• Add new document_cascade_metadata table with foreign key to documents (ON DELETE CASCADE)
• Add emdx/database/cascade.py module with cascade-specific operations
• Update documents.py to dual-write for backward compatibility
• Add migration_032 to create table and backfill existing data
• Update cascade commands and UI to use new module

New Module API

from emdx.database import cascade as cascade_db

cascade_db.get_cascade_metadata(doc_id)
cascade_db.update_cascade_stage(doc_id, stage)
cascade_db.update_cascade_pr_url(doc_id, pr_url)
cascade_db.get_oldest_at_stage(stage)
cascade_db.list_documents_at_stage(stage)
cascade_db.count_documents_at_stage(stage)
cascade_db.get_cascade_stats()
cascade_db.save_document_to_cascade(title, content, stage, ...)

Testing

• 22 new tests covering all cascade metadata operations
• Tests migration, CRUD, stage queries, and foreign key cascade delete

🤖 Generated with Claude Code