It's a long time since I looked at this, but IMO, the 'U-rod' doesn't stretch, but it's the location of the bend that creeps. The repetitively applied tension on the rod 'pulls' it around the pin. Compare the stretched compensator (left below) against new and it's clear that the first stage of failure is the opening up of the U-bend. This can only hapen if the rod is being forced around the pin. The rod becomes mis-shaped, so instead of being parallel, the two arms open out to the maximum angle possible within the confines of the 'case'. This reduces the rigidity enough for the rod eventually to get pulled around the bend until the increased compensator length prevents handbrake adjustment.
Sorry about the size of this image, I simply copy/pasted it from a 2011 thread
This explanation is proven by the fact that the very first method of fixing the problem was to simply weld a link across the ends of the rod. This stopped the 'stretching'. The substitution of the welding fix by using the steel shackle is mechanicaly identical. The two ends are anchored in place, so the effective length of the compensator remains fixed.
TC