Support for local LLMs

[Kurizek27]

Problem

being able to use huggingface/ollama/.... instead of online paid models

Proposed solution

access to ai without needing to pay for online models/without having to send possible private data to untrustworthy server farm

Alternatives considered

[uright008]

openai-compatible api as well plz

[mvanhorn]

Addressed by #16. Local LLMs via Ollama now work via ollama_chat/ (no API key needed). Hugging Face Inference also works via the LiteLLM huggingface/ prefix.