Why do you think this is for the flat position? If all you're doing is raising the front wheel, the attachment points on the bike (bars saddle, pedals) remain in a fixed position relative to each other so body position is not changed by lifting the front wheel. The rider might change position voluntarily with the lifted wheel, but that position is also available with the bike level. Further, unlike on the road, on the trainer there is no change in inertial load between the bike level and front wheel raised so the neuromuscular demands do not change with bike angle. Finally, since all the resistance to pedaling comes through the chain (gravity is balanced between the legs and pedals) tangent to the chainring, the direction of force from the chain also does not change as the bike is rotated. So it seems all lifting the front wheel does is change the direction of the gravity vector which might change how the rider feels on the bike and the weight distribution between hands and saddle but does not change the resistance forces on the wheel or the muscle recruitment needed for pedaling. That being the case, why such a large difference in power?

Bump. Good to see this thread back with new information. I still have an academic interest in this topic. Fun to read through the old posts. I'd forgotten all about this. My guess about the pro girl would be that her coach allows her to open her hip angle in the climbing simulation.

My spin class instructor has us do a fair bit of work at high and low cadences in TT position and similarly a lot of flat back work out of the saddle. Off the road like that, I've been able to work on my muscular recruitment in those positions and cadences. It's interesting how much one's pedaling mechanics have to change with changes in rider position as well as cadence.