AI03 - Improve Retrieval Flow for chat_with_rag

[Elly168]

This PR improves the retrieval flow for chat_with_rag to enhance RAG response quality.

Changes made

• Improved retrieval flow to return more relevant context
• Reduced duplicate and low-quality retrieved chunks
• Made retrieval results more consistent for user queries
• Added notebook to test and validate retrieval improvements

Purpose

This supports AI03 by improving how context is retrieved before generating responses, leading to better and more accurate outputs.

[Liam301020]

Tested locally. The notebook-based retrieval validation works as expected and the sample outputs look more relevant. Based on the current diff, this PR appears to demonstrate retrieval testing/validation rather than a direct runtime update to /ai-model/chatbot/chat_with_rag.

[Liam301020]

The current notebook changes are useful for validation, but AI03 requires improvement in the active RAG service / retrieval flow itself. Based on the current diff, additional runtime retrieval changes may still be needed to fully satisfy the task scope and DoD.