Yeah, I read over at the Castle site that this could happen (output v = input v). I'm surprised your receiver made it ok.
IMO, knowing about this issue, Castle should've incorporated "crowbar protection" - which has the effect of dropping a crowbar across the power line. In the event the output rises too high (~10v to allow for the full 9v range) a device would short the output. This would obviously be "bad", but the circuit is bad anyway to allow the full voltage through. Of course, shorting the output would burn up the BEC and/or the small wires, but sure is better than burning up expensive servos/rx/fans/etc. I think an SCR device and a couple of support components would be all thats needed.
A little explanation:
http://en.wikibooks.org/wiki/Practic...rowbar_circuit
You might argue that better circuits could be created to deal with this, which is true, but this would be an add-on part...