I see the same problem with a highres clothplane and V4s head. It looks like the original position of the head is not updated during calculation. It does change somewhat if you lower the offsets and  friction, but it will never settle the right way.

Looks like a bug here. Are you going to report this to Smith Micro?