Sorry; English. :)
I think it’s excellent to implement this in the correct manner and luckily I did it like that in the past as well. But it will prevent a lot of DSLR users, who did it in the wrong way, from processing old data right? The flat calibration is done, but visually it seems it isn’t. That creates the confusion, I had something similar in a different situation. So looking forward for very clear warnings. So in the end these people just have to use PI or something to use the old flats then, which is fine by me, but I can imagine it can create some disappointment for others. So maybe doing it the PI way with HUGE warnings and telling people they should change their workflow in the future might be a good thing for a first version? I however totally get you don’t want to get into a slippery slope. ;)