Feature request: improve UX for connecting to existing llama-swap/llama-server

[lostmsu]

llama-swap provides both llama-server and Open-AI compatible models endpoint. It would be nice if llama.vscode during initial setup asked if I already have server-compatible endpoint running, and checked if it also supports models request.

[igardev]

@lostmsu Yes, makes sense to facilitate the setup of the models for the user. Thank you for the idea, I will think about it. There are 4 models, which could be used - for completions (obligatory llama-server), chat, tools, embeddings and the user might not need all of them. For now the approach is when the user needs a particular model and it is not configured, llama-vscode asks the user to configure it and shows the UI for selecting a model. In this UI there is also a More button, which could be used to add new models.

I am not sure how llama-swap works, but if it offers one endpoint, which could serve different models (configured by llama-swap) I guess you could use it if you add this endpoint from button More -> Add external ... model (without setting model).