Feat: Extend available LLMs

[tanmaygupta9]

Extended llm.py and relevant configs to accept Anthropic's Claude-3-Opus and Google's Gemini-1.0-Pro models.

TODO:

• Test more prompts, particularly with Gemini, which might suggest additions for improving performance
• Fix streaming issue in Gemini
• Update .env.example to include ANTHROPIC_API_KEY and GOOGLE_API_KEY
• Add docs on how to add/test new models

[tanmaygupta9]

@kaikun213 forcing it to use gpt-4's tokenizer here through tiktoken because otherwise langchain tries to import claude's tokenizer from transformers otherwise (and I don't think we want to add that to dependencies). Does that work?