Add ability to use models through llama.cpp

[Flux159]

Ideally through https://github.com/withcatai/node-llama-cpp node bindings, although if openai support is already added, then maybe just use the api exposed via llama cpp's server.