Interesting and strange problem.
It seems like you have already done a great deal of smart work to it. Most of the first guesses I would be thinking about you have now addressed. Good idea on the replacement of all the sprockets and the work to ensure the idler sits straight.
The machining of the oil pump must have been quite a job! I assume you had to take the whole pump assy off the engine first?
Correct that the original type water pumps had flanges to guide the belt. In fact the pumps available here in the US continue to have those, so they are still possible to obtain. However, I don't think they alone would correct the issue you are having, since the belt should ride square on the sprockets even without them. To my recollection there are guide flanges on the crankshaft sprocket anyway.
I would be happy to send you one of the water pumps with guide flanges on the pulley if you would like -- I have a few NOS -- but again I'm not sure it would make a difference. Even so, perhaps worth trying.
The rear cam cap does have the thrust surfaces for the cam so I think your observations about axial play in the cam with that cap removed makes sense.
I am looking back through the earlier discussion now on this and I realize there is an important question that I don't think has been asked yet: do you remember if this problem existed already when you got the car, when it had the belt it came with? Or did the belt riding off kilter only begin after you did the timing belt replacement?
That may tell us something about my final guess. If the previous belt rode square but this belt now is riding offset, then it means the phenomenon must be caused by something in the timing belt components, as you have pursued so far. BUT if the other, old belt was already riding off kilter, then it could be due to the installed position of the cylinder head.
I have seen it be the case that after a cyl head R&R on these engines, the timing belt's position on the sprockets can change. There is just enough free movement possible in the headbolt holes in the head that the head can be not perfectly parallel to the block and I believe that causes this to occur, since the cam sprocket is not sitting square to the rest of the components.
That may be the only explanation left given all you have checked out so far. If that is indeed the cause, then the only solution might be to R&R the head, but that's honestly quite a bit easier than the oil pan job you already did.
