I just spent some time searching for a VW/Audi TSB that I know I have found before but cannot seem to track down right now. However this thread makes mention of it.
https://forums.tdiclub.com/index.php...ng-off.373906/
I went through something like this on a TDI engine once a couple of years ago. Per the instructions in the TSB (which are summarized in the thread above in the link), I had to remove the injection pump and reposition its bracket. The way it had to be repositioned seemed counterintuitive for the direction I wanted the belt tracking to move. But it worked.
I'm going to keep looking for the TSB but you get the idea. Of course in the case of the Volvo D24TIC engine, where the IP is driven off the rear belt not the front, this exact procedure is not relevant. But the general principle is that even just one component that is very slightly out of parallel can have the effect of changing the belt's position. (And can lead to damage of course as happened to the fellow in the thread above)
One thing I have realized in reading your last post is that you have a late style D24TIC engine with a dedicated tensioner for the front timing belt and a fixed position water pump. Thus, my earlier comments about the availability of pumps with flanged pulleys are irrelevant I now see, since those are the old style pumps with slotted mounting holes -- not what you need.
I'm not sure on your question of whether the back of the idler roller should be flush with the water pump pulley and in any case I am not sure my examples would be the same as yours, given that they use the old style system with slotted WP. However I don't see how making that pulley narrower will make any helpful difference with the belt's position.
THE GOOD NEWS, despite all of that, IMO, is that this is a pretty simple system.
If my understanding of the late style system is correct, there still are only a total of 5 components the belt touches: crankshaft, camshaft, water pump, idler roller, and tensioner roller.
1. OK, so based on your update, we now know that this problem persists with two different water pumps with different pulley sizes. That rules out a problem with the WP, I think, unless it is something really sneaky like the face of the block where the WP mounts being worn somehow to where it is not square. Highly unlikely.
2. You had the idler recess sleeved already and set right square to the pump mating surface. That probably rules the idler out as a possible cause. Probably.
3. The tensioner roller -- have you replaced the stud that it rides on ? Any chance that stud could be bent?
4. Crankshaft. This of course is the one part that cannot move.
We'll assume the problem is not here especially since you replaced the crankshaft gear.
5. Camshaft. This itself cannot move, but the cylinder head could be positioned imperfectly on the block. As I mentioned before, this is my #1 hunch cause. The reason I think so is that I have personally seen the effect where a timing belt rides in a different location on the pulleys after a headgasket replacement job. In the cases I have seen, it was not far enough of a change to result in the belt riding off the pulleys or to cause any serious problem. It was just different but still an acceptable position. However, I think a more extreme version of this could cause what you have seen.
Don't give up now after putting in all the effort you have done -- no doubt there is a solution. I would probably try replacing that tensioner stud first, see what that does, and if no luck there, I would pull the head and see if you can get it lined up better. A fresh headgasket and opportunity for a valve job is a nice side benefit.
