I only skimmed over the new posts, so please excuse me if I don’t perfectly represent some parts or miss others. I just have an urge to comment on the “maintainers this, maintainers that” parts:
I believe maintainers do consider this important, and I know at least one does (I). However that doesn’t really come into it, whether or not something like selective sync happens or not doesn’t depend on what maintainers think of this (beyond that basic “yes, the feature is welcome” stance that you’ll find all over already for this). We are all full-time employed (not to work on syncthing) and we need to deal with all things syncthing (that’s what maintainer means). There’s simply not much time left for us to “do things”, especially not big changes. And besides the churn work, there are also a crapton of “very important” things, this is just one of them. And yes, we tend to prioritise issues both by our goals and our interests/abilities, where the latter admittedly doesn’t lie on the front-end for any of us. So it’s accurate to say that you shouldn’t hold your breath for us to implement this. However there’s already one third-party tool that does it, one beta ios app that does some of it, one alpha quality abandoned attempt at an android app (and java/kotlin BEP implementation), multiple PRs attempting to add it to our UI (remember at least 2, one of which quite recent), … What I am saying is that syncthing and the ignore system functions in a way, that all the pieces are already there for some kind of selective sync feature, and we maintainers will definitely not stay in the way of it, the contrary.
Oh wow, typically for me I just wanted to write something short about “maintainers”, and end up with a wall of text.
tl;dr:
Don’t focus on maintainers. We will not stand in the way of getting this feature done (the opposite), syncthing already has pieces in place to make some variant of it possible and we’ll keep working to make sure syncthing stays usable (and thus ready for this feature) - indeed for a variety of reasons it’s unlikely that we’ll implement the feature.
(Damn, even my tl;dr is a classic German-style bandwagon of a sentence (well two). 
)