Jarvis: AI note-taking assistant

• models
  • new: Claude 3.5 and 3.7, and support for custom models (OpenAI-compatible endpoints) (FR #34)
  • new: Gemini 2.0 models (flash, flash-lite, pro)
  • new: GPT-4.5, o1, and support for other custom o-models (OpenAI-compatible endpoints)
  • deprecated legacy models: GPT-3.5 (still available as a custom model), Gemini 1
  • improved: removed constraint on max_tokens, memory_tokens, notes_context_tokens
    • previously, max_tokens, memory_tokens and notes_context_tokens were tied together due to how models used to calculate their tokens
    • many models today separate context, or input prompt, from the length of the generated output / response
    • so it is now up to the user to ensure that the settings match the model's limits
  • improved: model descriptions, separating input tokens from output tokens
• new: cancel button for database updates
• new: setting: Notes: Abort on error (FR #39)
• improved: model error handling
• improved: reset default API key to empty strings (FR #33)
• imporved USE model follows rate limit settings
• improved: UI style

Full Changelog: https://github.com/alondmnt/joplin-plugin-jarvis/compare/v0.9.1...v0.10.0