Cockpit reloading when the vehicle address or the WebRTC address changes is way less than ideal

We made them that way to be able to make them work without stopping to rewrite much stuff, but it's very annoying and for the user may seen broken.

The issue names no files, tests, or entry points. Start by locating where vehicle and WebRTC address changes are handled and where those changes trigger reloading; determine the intended behavior before changing it. Done means address changes no longer make Cockpit appear broken by unnecessarily reloading.

Written by the indexing model from the issue text.