I would expect that this model is informed by the several species of bat that use this method, and by the size and structure of the limb bones, which would be a

I would expect that this model is informed by the several species of bat that use this method, and by the size and structure of the limb bones, which would be a clue to the forces being used. If the animal can just get itself gliding, I think it is possible to continue generating lift even while raising its wings into position for a downstroke - what it needs to do is maintain a positive angle of attack while doing so. At least when they are not hovering, birds continue to generate lift on the upstroke. A headwind would make it easier, and perhaps that or a downslope might have been necessary.