feat: add Claude + Oracle AI Database Vector Search RAG cookbook example

[Ela689]

What this PR adds

This PR adds a new third-party cookbook example for building an end-to-end Retrieval-Augmented Generation (RAG) pipeline using:

• Claude (Anthropic) for answer generation
• Oracle AI Database Vector Search for retrieval

The notebook is designed as a practical, runnable walkthrough and goes beyond a minimal demo by showcasing Oracle-native vector capabilities in a realistic RAG flow.

Highlights of the example

This cookbook demonstrates:

• Oracle native vector storage using VECTOR(1024, FLOAT32)
• Real similarity scoring in the database via VECTOR_DISTANCE
• Hybrid search (vector similarity + relational SQL filters, e.g. category)
• HNSW vector indexing for scalable approximate nearest neighbor search
• End-to-end RAG flow: ingest → embed → store → retrieve → generate with Claude

What’s included

• third_party/OracleAI/rag_using_oracle_ai_db.ipynb
• third_party/OracleAI/Readme.md

The notebook covers:

1. Dependency installation
2. Environment variable setup (.env)
3. Oracle AI Database connection (oracledb)
4. Vector table creation with relational metadata
5. Embedding generation and document ingestion
6. Similarity search with real distance scores
7. Hybrid queries (vector + SQL filters)
8. Grounded answer generation with Claude

Environment support

The example is written to support both:

• Oracle Autonomous Database (recommended for cloud demos / production)
• Oracle AI Database Free (local) via Docker/Podman

Both use the same Python oracledb client API, with only connection configuration changes.

Why this is valuable

This example helps users understand how to build a production-oriented RAG pipeline with Oracle AI Database Vector Search while preserving the strengths of relational SQL (metadata filtering, structured queries) alongside vector retrieval.

Testing status

• Notebook tested end-to-end with Oracle AI Database Vector Search
• Retrieval and generation flow executed successfully
• README and notebook content updated based on review feedback

Happy to iterate on any additional feedback. 🙌