Re: [1] (I took the liberty of inserting numbers for easy reference), I agree with your observation about the time lag, but that should not preclude creating the PR for Gramps core and I would encourage that in this case.
Implementing @dsblank’s suggestion of moving to faster releases, and even nightly builds would alleviate the lag between PR and release.
If you’re proposing to test a fix/enhancement as an addon first (if it were indeed possible) and then get it ready for Gramps core, feels like a lot of effort, and will require two implementations. I would not be in favor. Apart from the effort issue, back-porting a fix to older builds as an addon (when possible) is a nice strategy!
Ideally, with the pace of enhancements coming in Gramps 6.x, I would imagine that our community would want to move to the newer releases rather than stay with the old even though I understand some people may hold out for various reasons.