New doc - Build a local RAG app

[alexwolfmsft]

Describe the issue or suggestion

Add a new doc to the .NET AI docs to cover a minimal RAG scenario using fully local tooling and AI models. Current RAG coverage relies on cloud services. Include the following:

• How to work with local AI models using Ollama, such as phi3 or phi4, embdding models
• Use local vector storage/databases such as qdrant

---

Associated WorkItem - 368771