Anthropic models, styling and comprehensive refactor

[gregriff]

Anthropic Models

Added support for newest Claude models:

• Invoke and interact with them exactly the same as OpenAI models
• Using official token counting/pricing method
• With OOP refactor, adding new models in the future will be easy

Refactor

Made many passes to refactor all parts of the program. Main focus was on OOP principals to make codebase easier to understand, maintain and extend.

• Abstract LLM parent class with OpenAI and Anthropic as subclasses
• Moved many impl details into classes, making main logic cleaner

Styling

Decided that easy visual customization by the user is very important, and a main goal of this project. I noticed that most similar projects do not do this at all.

• Added styling.py file to allow users to easily change the appearance of most things in the CLI. This is a precursor to proper configuration management with TOML or PKL.
• Adjusted cost displays of the program to align with goals of minimal UI/UX.
• Added loading spinner while waiting for first response

Misc features

• updated CLI arguments, options, behavior:
  • specify model name with argv[1] with fuzzy string matching
  • unix pipe support
  • flags added for all options
  • max-tokens option. controls max length of every response
• use escape codes to set window title, adapt to window resizing, disable keyboard input during output
• add multiline mode

Other

• updated all packages
• cleanups/removal of unness. code everywhere
• add credit to github project that I got inspo/implementations from
• README updates to better describe goals and show distinguishing features