Add image support to Chat and VaultQA mode

[Bugsia]

Most of the modern LLM's support images and, for example, with the Google API you can pass them as base64 encoded strings. So it would be nice, if you could do also do this here, and for example ask something about an image and then the AI answers with information form the vault.

I have tried to implement this myself but I fail to understand, how I can add an extra JSON field to the request body. Maybe someone can give me a pointer here.

Also maybe you have planned something like this for copilot plus but as far as I understand the code, it, at least partially, runs of you're servers and this alternative would work with the api key of the user.

[jsrdcht]

I remember seeing a similar feature request a few months ago, and I left a comment in support of it. I continue to support it now. Understanding the images in the documentation is very important.