Add Ollama as an Embedding Provider

[PratikDavidson]

Description

This PR introduces Ollama as an alternative embedding provider to Hugging Face, enabling local CPU/GPU embedding inference, enhancing data privacy, and reducing cloud dependency and cost. It also improves flexibility by allowing users to choose between cloud-based (Hugging Face) and self-hosted (Ollama) embedding generation.

Changes

• Added ollama client dependency to pyproject.toml.
• Implemented Ollama configuration and centralized error handling in config.py.
• Added model definitions for Ollama embedding models and extended EmbeddingService to process Ollama-based embedding requests in embeddings.py.
• Introduced unit tests for Ollama integration via TestEmbeddingServiceOllama in test_embedding_service.py.
• Updated README.md with documentation on using Ollama as an embedding provider.

Implementation Details

• Run Ollama server
• Config .env with:
  • EMBEDDING_PROVIDER=ollama
  • OLLAMA_HOST=localhost
  • OLLAMA_PORT=11434
  • OLLAMA_MODEL="nomic-embed-text"
• uv run server.py

Checklist

✅Code compiles successfully
✅Unit tests added and passing
✅Documentation updated
✅No breaking changes introduced

[nicholasericksen]

I was also looking for this capability. Thanks!