Good point. For those still confused about where the torque "goes".
Another way to think about it is this. You have a torque wrench and you're turning a bolt. Just because you know that you're able to exert 150 foot pounds of torque, doesn't mean that you do when the nut is spinning freely - it's just taking you a lot less effort (i.e. energy). Only once the bolt is tight are you able to exert the full 150 ft lbs. A spinning wheel is like the bolt before it's tight. Likewise, undoing a bolt is like the wheels coming off the ground. Once the bolt loosens you need a lot less ofrce (torque) to spin it - or if you apply the same amount of force you spin it much faster (like an air ratchet or the like).
Power is just an expression of how fast work can get done - the amount of work being done divided by the time it takes to do so - so with the viscous couplers, when one drive axle breaks free more power is sent to the axle with traction since it takes more force to turn those tires. Since it takes more force, more work is being done at that end over the same period of time and hence more power.
For the creation vs. destryed idea - remember that the energy we're talking about is coming from burning gasoline, so as was mentioned, if you kept applying the same amount of gas (energy) in neutral or on ice, you'd spin things faster since the same amount of energy is translated into rotational motion - so, you let off the gas.
I'm beginning to digress, so I'll quit with one last thought. Torque does not 'require' friction for a given value, it's like straight force. If you try to twist an anchored wooden pole with a lever, even after you've twisted it some and stopped, you still need to apply a force to keep your lever from springing back at you. No movemennt is involved, so there's no friction, bbut you're still applying torque.