The only reason that placement of the magnet is critical is because it must pass very close to the sensor which will be mounted either on the chainstay or seatstay. Likewise the cadence magnets position on the crankarm must be precisely located so as to pass as close to the cadence sensor as possible. All computers that measure speed do so with a magnet and sensor and those that add cadence must have two magnets and sensors, one for speed and another for cadence. High end computers can extapolate gear ratio's from the cadence and speed sensors together but cannot extrapolate speed from cadence. Using the GPS to calculate speed adds a new twist and I am not certain that it is an improvement, I can imagine certain kinds of riding and locations that would throw a GPS trying to deduce ones speed that would not faze a standard magnet sensor at all. I doubt very much however that the Edge will rely soley on the GPS to provide speed data. I could certainly be wrong of course but IMO it would be one more reason to think about a more general purpose GPS optimized for that task and a decent cyclecomputer likewise appointed.
H