In fact, the net load on the thrust bearing may be forward at one operating condition, and aft at some other condition. If such a crossover occurs, it's less than desirable; it causes chatter in the lightly-loaded thrust bearing. Designers will make every effort to move the crossover to some sub-idle point.
Hmm, would the sub-idle crossover for a typical Low rotor spool (GE & PW) start off with the bearing load aft, then switch over to forward. While the high spool would typically start out forward loaded and stay forward even at windmill?
It seems to me that you could always plan on a pressure rise across the high comp even at windmill and might even bleed some of this to a low rotor cavity to ensure that thrust bearing doesn't float when you don't want it.
But at some sub-idle there ain't any pressure to fool with.