Jgwill-odaia

- Upgraded Llama-index from the legacy version `0.9` to the current version `0.10` (note: upgrading PyPi version of PyGPT to `2.1.0` requires reinstallation in a virtual environment to remove old library versions).
- Upgraded Langchain to `0.1.9`.
- Upgraded PyInstaller (and compiled versions) to version `6`.
- Removed LlamaHub and restored support for Python `3.12`.
- Fixed imports for Llama-index in the compiled versions.
- Added icons for quickly toggling audio input/output.
- UI fixes.

- Added core for loading web resources (not only webpages) to Llama-index.
- Added a data loader for transcriptions of YouTube videos (simply ask for index a YouTube video URL using the `Command: Web Search` plugin - the `web_index` command must be enabled).
- Fixed the "Initial token count exceeds token limit" error in Chat with files mode.
- Added missing translations for dictionary items in settings dialogs.

- Added "simplified" mode to the "Audio Input" plugin, now set as default, allowing for faster and simpler speech recognition without requiring any configuration. Meanwhile, all current options have been moved to "advanced" mode.
- Improved microphone support in Snap version.
- Improved file browser update after performing file operations.
- Added check for file availability for indexing in the file browser.

- Improved and refactored debugging

- Added auto index file on read option in the `Files I/O` plugin
- Improved debugger
- Fixed Bing speech recognition

- Fixed empty API key dialog.