LLM chatbot that uses Blawx encodings as a tool

[Gauntlet173]

I've done some initial experiments on getting langchain and the -0613 models from OpenAI to generate JSON for the interview endpoint and use Blawx as a tool for answering questions. The basic concept works, but there's more to do before adding it as a feature to the app:

• verify that we can get it to generate true, false, and unknown facts
• verify that we can get it to generate ground and unground facts
• figure out how to get the chatbot to only use Blawx, and refuse to answer questions it doesn't have an encoding for
• figure out how to give the user access to the explanations generated by Blawx inside the chatbot interface
• figure out how to re-implement the gradio-style chatbot interface inside the web app