Smaller forelimbs were retained in most dinosaurs, including the quadrupedal ones (the obligate quadrupeds did have lengthened forelimbs, but with some notable exceptions, the hindlimbs were always longer than the forelimbs).
Dinosaurs essentially started out bipedal, so there was no particular reason to reduce the forelimbs on that basis. Maintaining the center of gravity over the hindlimbs is certainly important for bipeds, but it still doesn’t seem likely that such was the reason for the drastic reduction in animals like Tyrannosaurus and Mononykus.
I did specify the evolution of dinosaurs, so if dinosaurs started out bipedal, they descended from some animal that did not. Nobody knows why dinosaur limbs evolved the way they did, but better balance would be a clear advantage. However it happened, dinosaur genertics had clearly displaced genes that favored similar fore and hind limb sizes, so variation in limb size would not be drastic, even in the cited cases.
Rather than providing a positive assertion of the balance theory, I would say that it makes more sense than the grasping or standing up theories. With the exception of grasping for intercourse, the shortened arms of T. Rex would have been useless for predation or feeding. They wouldn’t be strong enough to hold any prey that they could get close enough to grasp, and they are useless for feeding since they don’t seem to reach the mouth. As for standing up, an animal with such an enormous head, neck, tail, and legs would not need short weak arms to get to it’s feet.
Oh, and in case I wasn’t clear earlier, the cited examples all have larger heads or shorter tails that require an adjustment in forward weight loading to maintain balance. As you noted, the quadrapedal dinosaurs developed larger and heavier forelimbs.