Could be that as they attempted to recover with no power available the RRPM had decayed so that when the engine re-engaged the differential between the N2 speed and RRPM was well beyond normal and the overstresss ocurred
If N2 is rising from around 65% as throttle is re-opened, and RRPM has decayed the differential would be smaller wouldn't it? Slightly off topic - is the reason for opening the throttle over 5+ secs when recovering to prevent a shock on the sprag clutch when N2 & RRPM meet, or prevent compressor stall or excessive N1...? I don't think there is a large change in torque (as there would be on the ground) as the engine is not having to spin up the blades in auto at 100%.