Well chrome has been doing that for a while too but what I mean is more like significant parts of the url for trust like say the user in github.com/<user>/<project>, because the user represents a significant silo similar to if it were instead organized like <user>.github.com/<project>

For example if I were to go to something like https://raw.githubusercontent.com/<user>/<proj>/<path> or

the most relevant parts of the url as far as security/trust is concerned is the domain githubusercontent and <user> not the raw subdomain

Well, whilst that is a nice idea I can't see it happening anytime soon. Attempting to work out if a part of a URL represents a user name seems like a bit if an impossible task to me. I guess you could encode rules for specific well known sites but I doubt you could ever create a general solution, and if a site changed its URL you would become unstuck until you rolled out a fix.