docs(rag): Add usage guide for retrieval logic

[erikleblanc]

Write a short doc about the retrieval flow: from a chat prompt, to embeddings, to an LLM that uses the returned context.

Wait for

#37 (Indexing API) and #38 (Document upload) so you understand the entire pipeline.
#39 (Indexing workflow) possibly, so you can reference it.

Next Steps

Potential integration docs for the ChatService that calls the RAG service.

Checklist

• Clear description of steps: prompt → embedding → retrieval → re-injection into LLM.
• Where logs or error messages appear if something fails.