What happened here????

[DKS]

I would guess dust explosion. The roof buckling is too uniform for a shifting load. The Sides show no deformation, crumpling.

Ah, but a dust explosion ought to have deformed the sides as well, no? Everything should be bowed out; but the sides are bowed in. Look carefully at the roof; the deformations form somewhat angular kinks, particularly toward the right end, as if struck by squarish-shapes. My guess is the load was heavy crates that only partially filled the car vertically. When the car tipped over (likely onto the side facing away from us, explaining the lack of crumpling on the visible side), the low load had room to gain velocity and slammed into the underside of the roof. This would naturally have the tendency to pull the sides inward in the process.

[AlkemScaleModels]

Depends on the amount of over pressure and the relative strength of the components of pressure loading. I would guess that the roof was weakest, then the doors and finally the walls. Thus when the blast popped the roof buckled first, then the door popped relieving the pressure, before the walls yielded.

[Robbman]

If you look carefully, you can see the other side of the car is basically bent well over halfway inwards... I can see where you're going in that the roof gave and took the other side with it, but the lack of any damage to one side and damage to the other shows a clear lack of uniformity, not to mention there's no damage to the door tracks, which tells me the sliding door wasn't removed by force  (the plug door could be welded, so I'm leaving it out of the picture))

A stamped roof is going to be stronger than those welded plate sides...