Choosing what to fix in 2.6 will be contentious, people will inevitably see a problem that affects them as vital, n matter that it only affects them. New DSLRs will be a particular issue.
The approach I'd take as a developer would be a triage one where I match the cost of doing a fix with it's importance. I'd expect something that affected few people, such as something hardware specific, not to be fixed but something that affected everyone, such as a bug in the core meridian handling process, to be fixed, even if a lot of work is involved. Things that are really easy to fix, if all that's required is to copy some files from the 3.0 to 2.6 sources, build and distribute, are more likely to be done.
All but the most trivial fixes will involve testing and maybe documentation, these are significant, testing particularly. A one line fix, taking 10 minutes to implement could take many hours to test, especially if it needs time under the stars.
One approach that may help is to do an occasional - maybe yearly - maintenance release.
BTW I don't see astro imaging as fundamentally expensive. Sure you can spend many thousands on the latest high end kit but you don't have to. A simpler set up can work well. Something like an AVX with an 80mm scope, a guide scope, a motor focuser and the family DSLR may not break the bank, especially if bought second hand. Once you have the kit the running expenses can be minimal.