feat: add agent_id isolation for multi-agent support
Add agent_id column to conversations table to enable isolation between different AI agents (ANIR and future Java agents).
Changes:
- Add agent_id column (VARCHAR 50, default 'ANIR') to conversations table
- Create composite index (user_uuid, agent_id) for fast agent-specific queries
- Add agent_id validation in ConversationService.get_conversation()
- Update all conversation service methods to accept agent_id parameter
- Update message_service to pass agent_id when accessing conversations
- Add logger import to chat_api.py
- Add Integration/scripts/ to .gitignore (development scripts only)
Architecture:
- ANIR (Python/FastAPI) as central memory service
- Other agents (Java/Spring Boot) calling via HTTP bridge pattern
- Database-level isolation prevents cross-agent access
All tests verified (4/4): ✓ Conversation creation with agent_id isolation ✓ Memory endpoint access (HTTP 200) ✓ agent_id column presence in database ✓ Composite index creation