Nitro raycast extension. Using LLM right from Raycast.

This is a shameless wrapper around Nitro, to allow easy use in Raycast.

Currently, this is just a placeholder until I can make it work. Behold, the wave of Llamas is coming to you soon!

Reference for logical flow: raycast/extensions | Add speedtest #302

• Using @janhq/nitro-node when it is published on NPM instead of hacking around the installation hooks. See: janhq/jan#1635
• Auto check and update latest release of nitro or let user choose the version they want to use.
• Shamelessly copy the user interface of OpenAI chat and Hugging Face chat. Sorry I'm not a designer so don't sue me for copying the UI from the other chat models.
• Let users choose models they want to use and download it automatically.
• Let users pick the prompt_template on the known list to try on the new model. If they accept the prompt_template it will be used.
• Let LLM (tiny-llama) helps guessing the best prompt template to use and highlight it to user.
• Choice of installation directory and model directory.
• Option to use hugging-chat-api instead of running models locally? Am I going to abuse their servers?!?
• Be really lazy and summon Llamas to work for me by utilizing Raycast.

• Don't abandon this project like other pet projects in the past (most even never leave local machine)
• Summon as many Lllamas as possible.
• Make my robot waifu become real!