I think lateral roll centre migration is a fairly benign characteristic, and I'm struggling to think how it would really hurt the car, if it occurred in the absence of any other bad geometry.
The problem I have is that if you have two geometries, and one has lateral roll centre migration, and the other does not, then they will tend to differ in other respects, which I think will dominate.
For instance, looking at the WOB linkage proposed further down the page, moving the body mount one way o the other has no effect at zero roll, but if you roll the car then you'll get a jacking effect. But really that jacking is due to the effective change in roll centre height, not lateral location, in itself.
I agree that the two ends of the car should be designed to move in a similar fashion, generally, but my recent experience with GRCs leads me to believe that you can get away with murder.
Cheers
Greg Locock