Support local and custom models

[leo4life2]

Issue #165

Done:

• Interface changes for adding a model with HuggingFace model name
  • UI logic: put HuggingFace model name in "add model" text area (e.g. owner/modelName), click Add, select the model name in the dropdown, and start chatting.
• Loading & running model from HuggingFace

WIP:

• stream tokenizer decode minor bug
• model name persisting (will probably use localSession)
• support local models

[AlexCheema]

This needs to use HF_ENDPOINT environment variable which was just added

[AlexCheema]

This looks great.

What we'd really need here to get this merged is automatically sharding the model. The PyTorch implementation here does this automatically, you can see how they do it #139
That way any language model can be supported without having to explicitly write a sharded model implementation.