the only other clue i can offer up at this point would be the fact that the bearing was not pressed in the motor housing. it came right out so maybe a poor press fit added to this?
Bearings are quite often an interference fit on the shaft but close sliding (or transitional) fit in housing. Makes assembly easier, especially in production environments. Interference fits on both the shaft and housing make assembly a nightmare.
Frequently for smaller motors a spring/wave/Belleville washer is used inside one housing end to permit the bearing to move axially as the rotor shaft grows (from heat). That spring preload also tends to keep the bearing outer races from spinning in the housing bore.
If the bearing was a fairly sloppy fit in the housing there was very little resistance to shaft bending and all the movement would have been concentrated at the shaft bearing seat shoulder.
Andy's suspicion about a fatigue crack from shaft misalignment is looking pretty darn persuasive to me.
