That is a strange problem.
So do you mean that even when the stalled motor gets past the problem point and has power again that it still doesn't spring into life? Very strange, I don't see how being in a consist or not would affect this problem. In both cases the DCC system should be sending out comands that say the motor should be going at a given speed.
What I have done with my LCE, which also has two decoders, is to wire it all of the way through so that the power picked up at the front is connected to the power picked up at the back. I've also fitted a few power pick up wheelsets at intermediate points. While this may solve your problem I suspect in reality it would just be coving the problem up, because neither motor/decoder would loose power.