I've used QML before, but I appreciate you linking a sample for anyone who's curious.
Re: the matter at hand, it's cool if you don't consider that bloated. If I was using a C++ Qt codebase as the comparison point, I might think that too. However, if we're comparing it to a typical web codebase, that's an order of magnitude more dense. This also just doesn't matter in 2018, because QML could have the best documentation in the world and you'll still have trouble figuring out any issues as it doesn't have anywhere near the level of literature and/or (discoverable, not buried in some dead Linux project) open source code that the web platform does. Qt has a sexy problem, and every time it's brought up people try to claim it doesn't matter, but it does.
(There's also a slew of apps (e.g, I could go pick a cryptocurrency wallet app at random...) that have been built long after 2012 and still have issues with rendering.)
I'm not interested in debating KDE, since... that project stopped being relevant to me around KDE4. Nothing but wasted potential.
Re: respecting a Desktop theme... I used to love ricing out my desktop too. ;P
The thing is that it's an archaic viewpoint - one that you're welcome to have, but it'll increasingly be pushed down. Product design is always going to trump desktop UI customization capabilities, because one of those has money behind it and the other doesn't.
I also don't know anybody who customizes their desktop chrome anymore because (in addition to companies like Apple making it next to impossible) most people use their machines through a web browser at this point.
Also, let me be clear: I would love for native UI frameworks to be better. I don't particularly enjoy digging around in web codebases, but pretending that anything is else is as close to approachable as them makes it very difficult for any native UI toolkit to improve. People seem to think it's on the Web to catch up to the performance of native, but it's really the reverse - native needs to play the game how the web does or it's going to go extinct.