Roy mentioned in his rationale for making the switch now was that there were things he wanted to do in TBAK that simply wouldn't've possible if he had stayed in Daz. So it wasn't just a question of making renders faster. UE gives him better tools to create a higher quality product that better achieves his vision for the game.
It still feels largely like a creator mucking around with stuff they should have left well enough alone. I understand the urge to keep revising things, it's a problem I have myself, but there comes a time when you just have to put down the red pen and move on. Either work within the limitations of the tools you have, or head to a point that works well enough for a climax and make a part-2 exe that uses the new stuff, and you go back to tinkering when you're able to, rather than holding up the entire thing.
I'd probably be less skeptical about the decision if I knew that it was going to be for something of substance, rather than just making prettier renders. No disrespect to that aspect intended, it can be a huge pain to work out, I know. But it feels like shelving the story for literally over a year at this point just to make chapter 3 look nicer or whatever is a sign of misplaced priorities when the game has yet, to my knowledge, introduced any meaning to the various stats the game has, or given the player a chance to do anything besides continue down a very narrowly tailored pathway.
It still blows my mind that this game has a kidnapping plot (two actually) and several assassination attempts (and one success), but we still haven't really had any moments where we have to try and figure things out, or deduce if someone is lying, or anything like that.
About two years ago, I made a couple comments in this thread talking about the frustrations I have with the game's design, pacing, and priorities, and I haven't seen any real signs of improvement. Granted, that's because of a year apparently dedicated to making revisions, but I'm doubtful that those revisions are about anything beyond maintaining visual continuity between prior chapters and the future ones.
Would love to be proven wrong though.