Assembly isn't hard, given good tools. It's just a much different way of thinking/organizing code.
I'd wonder if that CPU "halt" that Maria can do for DMA may also cause some slowdown. If I understood correctly and my memory is still ok, I think the 7800 uses sprites for everything, even backgrounds? Meh, something to look into if I ever get time to do something fun like that.
This is absolutely true. The 7800 can do games like Robotron and Asteroids better than any other system of the time. No flicker at all. No limitations on the type or direction of movement. Rotating rocks that move really fast? No problem.
NES version of "Ballblazer" is a good one to consider...it just isn't as good on NES as it is on 5200, 8-bit, and 7800. Oh, and the C64 version is horrible.