feat: Add support for LLama models

- Add `--web_model` flag to `chat_gpt`, `askai`, `google_ai`, `claude_ai`, and `o_llama` commands to specify the model for summarizing web results.
- Add `--ollama` and `--ollama_model` flags to `ai_media-summary`, `ai_transcription-summary`, `ai_trans`, `ai_trans-sub`, `ai_debunk`, `ai_analyze_paper`, `ai_analyze_religious_text`, and `ai_fix-json` commands to use LLama models.
- Update `ai_tools.nu` to support LLama models in various commands.
- Update `config.nu` to add a new keybind for deleting one word backward.
- Fix a bug in google_ai where an error was not being returned when the API call failed.

Author: kurokirasama