SNS3Guppy as a ex mech eng it does tell you alot.
The intial crack is just outside the area of workhardening where they have folded the lip round just where the metal will have thinned due to stretching to form the lipped smoothed edge.And its bang on the stress concentration. Also its following the stress lines
A very simple example is shown below with a plate with a hole in it. Rotate it round by 90deg to get this case.
The actual model would be way more complicated because its thin plate and quite a complex 3D load it would be non linear as hell. I suspect if they have done any Finite element analysis on it they have fallen into the trap that they have done a linear model on it and ignored variations in loads due to changing airspeeds and engine loads.
What also gives me an indication that it was a fatigue crack followed by fast fracture is the fact that the paint has been blown off the intial crack. When the crack does propagate until the critical crack intensity is reached when it does go the crack extends at the speed of sound in the material. The sonic pulse radiates from the crack tip and because the material around the original crack can move out of plane of the rest of the body and vibrates in the (in this case radial plane) it pings the paint off because it is quite brittle.
And after looking at one today I would agree with the poster above that the bracket has failed on the base plate causing the failure.