

this is such an oversimplification. maybe it’s hard to distinguish between google services, but if you play some online game, chat over whatsapp or signal, or have a voip call, that’s an entirely different story. these can probably be told apart by DNS requests or active connections, and in the case of communications, messaging and voice calling is obvious to tell apart because of the difference in the volume of data. when having a voip call, through a service that supports peer to peer calls (most do, and it’s default on), an observer may even be able to deduct something about who you are speaking with, like what general area they live at.
then what if you have apps that try to establish connections to services at home. like smb or nfs, https services. your smb/nfs client may leak your credentials, I think even linux does not encrypt smb communication unless you request it in a mount option, and with HTTPS you leak your internal domain names because of TLS SNI.
Quite obviously the problem is not that you did not write an 560 page essay, but that you were misleading by basically saying “nah, it’s fine, nothing could leak, everything is ultra secure nowadays”.
did you just ignore a whole lot of points here? DNS, SNI? smb clients? whatever else? its not like I’m using HTTP. things are largely encrypted, the rest is out of reach!
how many sites exactly support that configuration? do you need additional configuration for that in e.g. nginx? if so, most selfhosters probably don’t have it, because it’s talked about almost nowhere.
and is it finally enabled by default in firefox? will firefox just retry without encryption when the connection fails?
it is certainly in scope. the discussion is not about security and your accounts getting hacked by evil EU, but privacy and data mining, for which all of these is a treasure trove.
probably not the coffee shop but the networking equipment, where even cheaper models include some form of “smart cloud security”