For example, I don't think that your murky pools will suddenly spawn fish after a hot-swap of raws post-worldgen. But that's just me and I could be completely wrong, but if I ask repeatedly "is it fully compatible with old saves", that would just get old fast.
I assume fish repopulation would be ignored for more natural processes, but it can be done, within limits. A civilization that starved to death from no fishing food source would not be restorable, even with a recorded deterministic history -- the player would not appreciate it when half his fort is spontaneously killed in a newly conceived and retconned war. It only requires a migration strategy, i.e.,
verX => verY. DBAs/MMOs do it all the time.
Toady One can only fix bugs that have been discovered. Which bugs are actually discovered depends on the style/habits of the players testing the game. So, if you don't ever play the game until you think all interesting bugs have been fixed, you may find that some bug spoils the game for you when you actually decide to play. And at that time, Toady One may have decided to move on towards the next release, making a bug fix something you have to wait months for instead of days or weeks.
A million software products exist that can use more QA helpers. DF is not more deserving. Concern over delayed fix inclusions presumes you care when DF reaches a favorable state. I can comfortably wait another 3 to 6 years, or simply never again play DF. The important part is knowing that Toady's vision will persist with new and exciting features, and that no amount of bugs, known or unknown, will prevent that.