You right, if you increase wire size then you can break even or come out ahead but at a higher cost. Less efficient use of materials so to speak.
For instance using an example of a 12/9 arrangement on an 8" disc with 40 turns. The coils on the 8" would be 7 inches per turn. Assume all other aspects didn't change such as magnets, air gap etc. Placing the same magnets on a 12" disc would increase the coil size to 10" per turn. The top and bottoms of the coils are wasted and there is still only 4" of wire in flux per coil. The 8" wastes 3" of wire where the 12" wastes 6" of wire.
It would be far more efficient to use 16/12 on the 12" disc. This would allow you to increase wire size, reduce the turns needed, ( both would increase output considerably ), as well as increasing magnetic area and wasting less wire.
As far as flux leakage, I believe it works both ways... If their to close you loose and if their to far apart you loose. So there is an optimum area but you can fudge a little either way and still come out good.
One last point... If the magnet has to travel a farther distance before encountering a coil leg then that is work wasted in my opinion. If the magnet leaves one coil leg and encounters the next it's more of a continuous load.
These are simply my theories on the subject based on my experience of designing and building literally 100's of them. I don't consider myself an expert on the subject but I have a reasonable understanding of what works and what doesn't, I'm still learning....