replace AsyncOpenAI with llama_index (WIP)

[jlzhou]

I haven't done complete tests on models apart from Ollama local models.

TODO:

1. Complete tests on openai and other remote LLM service(Need help)
2. Add llama_index dependencies into requirements.txt
3. Add extra model config params in settings
4. Update README.md to fit in with these changes(Need help)
5. You need to set OPENAI_API_KEY in environment with openai API

Suggestions and improvements are super welcome!

[jlzhou]

Also, if you're running ollama local embeddings model like nomic-embed-text, you'll need to change 'vector(1536)' to 'vector(768)' in two spots in the init.sql file, then execute it to create/re-create the db table.

[MFYDev]

Thank you for the commit, I will review this later this week and will keep you updated!