I remember a similar thing happening to the UWAM dyno motor when I was at uni. We were tuning the race motor and at about 2am we decided to turn it off and get some pies from the local servo. 30 minutes later we came back, turned the ignition on and BANG!! It was an impressive fireball and the only thing that stopped it from flying off and hitting the roof was all the sensor wires.

That manifold was made from carbon fibre and managed to stay together, but only because it blew off from the rubber joiners to the motor (Subaru manifold is bolted down and isn't going anywhere).

We could only guess at what caused it, but after looking at the inlet ports on the motor, you could see small pools of fuel sitting in there. So we just assumed that this fuel must have evaporated from the already hot engine into the manifold and as soon as we turned the ignition on, there must have been a spark inside to ignite this mixture. The resultant pressure build up inside the manifold had no where to go because the throttle is closed, so off goes the manifold from the joiners.

I guess the lesson we learnt from that was to make sure the manifold was earthed really well and it never happened again.