Testing is great. Challenge is "How" to test.
Interesting points by Concentric, but I fear that the focus on the spalling has led people to examine what we typically understand from spalling events on a bearing raceway, and that this event was NOT a typical spalling event due to ordinary Hertzian contact stress. Tell us more about the other fractures, and then we can talk about the likelihood of spall propagation to fatigue crack initiation and propagation happening on three separate regions at similar timelines. Maybe more likely that general rim fatigue began in multiple locations and the subsurface initiation produced a spall. Not the other way around.
So we rumor folks do not know the details of the ring stress loads on this planetary arrangement, maybe AH does but probably not. If you don't, then a test would also probably miss the load, and the test worthless in real benefit (Finding and correcting the problem). Most test stands lack something, even if just the ability to represent a larger load for test purposes. A piece-part test can do that, but then lacks the rest of the appropriate interface. A classic problem for engineers, not so easy.