Yes Tandem, that's been my gut feeling too, and you said it well. Thank you.
With the machine being at low level, if the MGB seized suddenly then it would have most likely departed the aircraft and the scene. If it seized slowly then the rotors would most likely still be turning during the short distance to the roof and there would have been evident rotor damage.
If the machine came down with an extremely high G landing you would also expect the rotors to slap down and end up slicing the throttle quadrant off the roof of the cabin and the tail boom off the back, damaging the rotors in the process.
But we cannot see any evidence of that.
So what if the roof is three layers thick, it's only designed to support itself, and like many old roofs it would be flush with many weak spots. 1 ton plus weight on each thin skid applied almost instantly to the roof and you can imagine it working like a knife to carve into these weak areas.
I feel the machine was skillfully controlled towards the landing spot on the roof, unfortunately the Fenestron shroud could have caught the aircon box and caused the nose to tilt down and roll the machine to the left with the left skid crunching into the front part of the roof first. The impacted roof could then have been prized from its supports from the front of the building, and with the masonry wall at the back being the only support left, the front of the roof eventually gave way and the machine fell into it.
As I said before, only those currently working at the scene of the accident are closer to the truth.