feat: provision + export documents table with generated SQL, SDK, and CLI

[pyramation]

Summary

Completes the provision → export → codegen pipeline for the documents table and all existing schemas. This PR builds on #34 (blueprint definitions) and adds:

1. DataId node restored — all schema files now include 'DataId' alongside 'DataTimestamps' in their nodes arrays. Removing ORG_NODES previously dropped DataId, which meant tables had no primary key constraint registered in metaschema. This caused chunk table FK generation to produce invalid SQL (REFERENCES parent_table () instead of REFERENCES parent_table (id)).

2. Removed duplicate BM25 indexes — SearchUnified already creates a BM25 index on embedding_text (which includes content via source_fields). The explicit content BM25 index on documents and notes was redundant and caused:

   • Duplicate filter/score fields in the GraphQL API (bm25Content + bm25EmbeddingText)
   • Double-counted content relevance in the composite searchScore

3. Provisioned all schemas — 11 blueprints (CRM, Agent, Runtime, Projects, Life OS, Email & Calendar, Staging, Autonomy, Documents, Cross-Relations, Spatial Relations) successfully applied to a live database.

4. Generated SQL migrations (packages/agentic-db/, packages/agentic-db-services/) — deterministic pgpm export from the provisioned database.

5. Regenerated SDK + CLI — GraphQL schema (1398 types), ORM client (95 tables including Document, DocumentsChunk, CompanyDocument, ProjectDocument), and CLI commands.

New tables added:

• documents — with SearchUnified (embedding + BM25 + chunks), repo/path/commit tracking
• documents_chunks — chunked embeddings for RAG (auto-generated by SearchUnified chunks: {})
• company_documents — M2M junction table
• project_documents — M2M junction table

Review & Testing Checklist for Human

• Verify documents table columns match expectations: title, content, metadata (jsonb), repo_name, file_path, commit_hash, tags (citext[])
• Confirm DataId is present in all schema files' nodes arrays
• Confirm only one BM25 index per table (on embedding_text, not on raw content)
• Deploy to a fresh database (pgpm docker start --recreate → deploy → provision) to verify migrations apply cleanly
• Check that the generated SDK includes Document, DocumentsChunk, CompanyDocument, ProjectDocument models with expected fields

Notes

• The alteration numbering in packages/agentic-db/ changed because DataId now creates the id column explicitly (previously it was implicit), which shifts the deterministic ID sequence for all tables.
• packages/agentic-db-services/ had some tables removed (apis, api_schemas, domains) — this appears to be a cleanup from the latest constructive-db platform.
  </pr_template>

Link to Devin session: https://app.devin.ai/sessions/2a4faca18f1a4b4aaf3d8e1162f67e0d
Requested by: @pyramation

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring