LLM provider, llm abstraction, and endpoints

[jspada200]

Create an abstraction to interact with an LLM. It should have one endpoint that lets you generate a response based on a chat thread. There are a lot of existing LLM abstractions that use the openai spec so these can be utilized. We may be able to forgo a provider entierlly and use one of these abstractions

Acceptance Criteria

• Able to use an endpoint to interact with one or more LLMs.