Thank you for the replies.
This happens all the way up the print, I've seen it 5-8cm up on prints which may seem to indicate the table is level. I've also spent much time leveling the bed so I hope it is quite square.
Based on this bit from gr5, "Try moving the print head and see if the belt is always straight or starts to head towards the pulley at a steeper angle as it gets closer." it likely has something to do with that. Thinking about the way this works, technically with a belt tensioner that raises the belt you will have a non linear plot of the belt speed at a constant pulley speed.
I added printed belt tensioners to all 4 blocks to make them properly tight in the past, this certainly changes the angle of approach as you get closer to the pulley on each end.
It seems, if I am thinking right, that this would exactly cause this problem. It also seems, thinking in this way, that any belt tensioner design which drives the belt upward and changes the angle of approach could be causing more harm than good.
I guess I need to get shorter belts that will be capable of being taught with the standard "tensioner" that pulls the belt down into the block and does not raise the edges causing this angle of approach problem.
I'm in Uruguay so getting belts is a slow process, but I will update this post with the results of this once finished.