Sure, it takes work and testing to get a UI into a condition that seems, to the end-user, to just work
BUT that kind of work, each in the small end-cases that only come up ocassionally, is needed. Sure "that doesn't affect a lot of users...".. yeah it will continue to "not effect a lot of users"* as long your usability sucks and not one uses the thing...
An app or desktop or whatever needs to fully implement the features it claims to have. And it is often true that the last yard turns out to cost as much as the rest put together. Yes but the last yard is necessary. I can understand if it won't get done 'cause there's not enough time or money. But the attitude that it shouldn't get done 'cause "that doesn't affect a lot of users" makes my blood boil. Sorry, I'm sure people put lots of effort into this stuff but I just have to get this off my chest.
This complaint is similar to desktop icons not maintaining their position when you change desktop resolution. It requires far more work than changing the way the backend handles shortcut keys (to fix the menu bug), and doesn't improve most people's experience.
I'm not saying you're wrong! This is a legitimate problem! I'm simply saying that it's not a priority, similar to how it's not a priority for Apple to expose a user interface to customize the button layout on their windowing engine, or fix their dock to work properly in vertical mode. All projects have priorities, open source and closed source, and this bug doesn't seem very important to me.
The problem with that thinking is that all of those little corner cases, in aggregate, create a situation in which you are constantly encountering little unpolished, sharp edged bits of ungainly behavior, and it's never going to go away until people reject CADT-type behavior and realize that polishing enough corner-cases for small subsets benefits large majorities in aggregate.
Shuttleworth seems cognizant of this with his support for the paper-cuts project, but there's too little of this polishing going on upstream for a 100-odd little fixes per-release effort to really tip the balance too far towards improvement.
Open source has a different problem from closed source in the sense that the "sausage making" is visible from the outside.
I don't think that processes or design principles of closed source companies and products is applicable. Is important to prioritize and manage what's happening ... It's just that it's important to do that in a way that makes lets the random person walking-in feel like they've begun to participate rather than giving an answer like "who let you in". (essentially should be the opposite of Linus managing the Kernel).
Perhaps a "friendly bug reception area" would be a worthwhile project. Launchpad may be very functional in ways but I'd say it's very, very broken as far feeling you're participation valued. A random bug thread is "comment-1, comment-2... little-icon-saying-this-is-not-a-bug-with-little-explanation". Just a longer boiler-plate explanation for changes like could be more soothing.
What I have experienced is the "Internet Interface" of ....
"Whatever the problem is, the first answer is 'it's not a priority'"
THAT needs to be tweaked. I understand prioritizing things. But we really, really, really need a way to relate to those frustrated by and reporting X that doesn't involve "answer 1: we don't care about X" because that has as it's implicit message "screw you".
Edit: Also, the original problem is something really looks like a bug to the end-user whatever is may look like the programmer. So fixing it should be "bug priority" whereas a number of the thing you mention are clearly features, look like features to the end-user and so should have "feature priority". Just consider...