Research if we can put the rendering visualization bytes logic from clientside to backend

[ddey2]

Currently, all the fall back logic if visualization data is not present is in client side (frontend). The code is a bit messy. Check if we can put that to a backend endpoint.

[max-zilla]

• visconfg can have parameters, but does NOT need to have vis data.
• for some components (e.g. thumbnail) we don't care where the bytes come from, either visData or raw bytes. so we want an endpoint to get the bytes for a given component - the backend can return from either.