fix : RAG health check uses global index instead of per-user index

[tmdeveloper007]

Summary of What Needs to be Done:
The GET /api/v1/rag/health endpoint calls check_index_exists() without a user_id parameter, checking only the global FAISS index path. In a multi-tenant deployment where each user has their own index under faiss_data/user_{id}/, this gives misleading availability information for individual tenants.

Changes that Need to be Made:

• backend/app/api/v1/rag.py (rag_health function): Change check_index_exists() to report general RAG module availability (e.g., embedding model reachable) rather than checking a specific user's index. Alternatively, remove the index existence check and instead verify the embedding model is functional.

Impact that it would Provide:

• Accurate health status reporting for the RAG module
• No false negatives when a user's index is legitimately empty
• Better observability for operations teams

Note: Please assign this issue to the tmdeveloper007 account.