Well, for one thing the LR process remember your previous class as there is for example a dialogue option that tell you which class you have been before. It is further not only a dialogue option but it serves the purpose of retaking the same class as before without wading through the other options. This is especially nice if you really want to fix a feat or like in GR want to go from 28pt to 32pt (which is by the way for what it was initially designed, not to help players to level an easy class and then swap them out at 20 to get a free PL)
Originally Posted by RobbinB
Additionally I believe the dialogues are integrated into the workflow and not just placed independently on top of this. In other words would need some refactoring which lead to potential bugs. Bugs on the LR process are however scary and a critical part which is why they probably don't want to do it in a rush.
Further they thought it probably simple, on the code level. So instead having an additional variable that recognize the previous setup, they keep the complete old character level structure, and the LR process just swap out the current level one by one, like a pointer that is shifted for each processed level.
So I can see why it is difficult for them to change that, but I would also like to see this limitations that it brings gone.