Actually, it's called IA. User interface needs top-down work. That's information architecture to improve the UI and the UX. With correct and well documented IA in the first place, UI overhauls shouldn't be required. However, I'm not here to argue semantics. But thanks for trying.
The information architecture is pretty good (it has been from the start), which makes it easy to extend the theme. The reason for the UI overhaul was a change in requirements from the theme review team and not a decision made on a whim by the author - every theme has to go through a review process and this theme is a lot older than the review team, hence big changes required by the review team for some reason generally necessitate UI redesigns. This is not the only theme affected by changing requirements: a lot of other featured themes face the effects.
It's clear the author has put an almost inordinate amount of work into this, but there needs to be some direction, I think. Often, when we work in isolation on personal projects, we lose sight of what is best for the users in favour of executing the myriad ideas we have bouncing around our heads.
There is a support forum with more than 15,000 posts - this has neither been directionless work nor has it been a "work in isolation on personal projects" for almost a year and a half now.
Floating confirmation dialogues are little more than a distraction, and I defy anyone other than the author himself to know with great confidence their way around the back end. There's simply too much to memorise.
The user base runs to more than a few thousand (the weekly downloads are upwards of 5000+ for most weeks, and the number of active sites with the theme probably run into 7 figures), and there are quite a few people who are pretty well versed with the back-end: not just the author. Also, quite literally thousands of users haven't had issues with the 1000px width, and for those who did, the theme doesn't lock you in at that width.
Most options and the way you see the UI are a result of user-feedback, so the design is pretty consistent with what most users want. Users do provide occasional feedback on the support forum for how they would like the back-end to be structured, and you are welcome to pitch in there with what you think is wrong.
Regarding the floating "save box", there is a plan to reduce it to a smaller button on the side, which would still be floating. Mind you - this earlier was a single row of buttons at the bottom/top of each page and that just didn't work because of the large number of options on each page: you would just keep scrolling endlessly to save changes.