Even LiFePo4 can spontaneously combust. They are more resilient, though. The goal is to reasonably minimize the chances of fire, which I think can be done either with the hybrid idea or a pure LiPo approach. I have bought some A123 cells to test the hybrid pack and see if it works.
Power is F*v, so while the force is m*g/(L/D), the drag losses are m*g/(L/D) * v. The v term is going to always be around 25m/s, almost independent of the glider.
Also consider that when a propeller pylon is deployed the drag increases tremendously. On the AC-5M (published 35:1 L/D), the ungainly engine pylon
is literally as much drag as the rest of the plane combined, so the L/D drops to 18:1. This means that drag power is doubled.
Fortunately, a motor has much less drag than an engine, so we can expect a bump in L/D. I'm conservatively penciling it in at 22:1.