> Distributions are biggest obstacle here: sure they solve a lot of other
> problems, but then basically build impenetrable wall between developers
> and users,
Which is just a convenient excuse. Don't drink the Wallyesque kool-aid, sometimes the developer complaining (should I say boasting?) about the 'impenetrable wall' is even the same person as the packager that pushed the problem version in the distribution.
Not looking at distro reports is 100% the developer choice (and other software projects, which are not even managed by people paid by a distro company, have no such problems).
Drowning under distro problem reports is 100% due to buggy or incomplete software that needs fixing. If you move users from distros to somewhere else the software won't be any less buggy or incomplete and the reports any less numerous.
The same upstream bugs that go unfixed for years are duped by Fedora people, then Suse people, then Ubuntu people, then Debian people, following the deployment calendar of the problem software versions in various distributions, you can see the waves of dupes that occur when an version is deployed by a new distro, and it has nothing to do with any particular distro and everything to do with plain unappealing software bugs.
Solid projects are *happy* to see their software used in many different contexts since that helps identify difficult-to-trigger bugs (see this year's FOSDEM LibreOffice presentation, the many times Linus stated how happy he is to see the same kernel used on many archs and from embedded to big-iron, etc).
But it is easier to shoot the messenger than to fix problems.