I could also attach an ANT+ speed sensor to that bike to measure speed and distance at the wheel instead of relying on GPS.

My concern with using the dynohub wheel for testing the Magnics is that it's not really fair to the Magnics to add the (admittedly tiny) amount of drag that the dynamo imparts. One way to account for that would be to perform four tests -- split the Magnic test into one with the dynamo wheel and one with the non-dynamo wheel.

You're likely right in that this wouldn't be able to reliably detect any real drag difference between the two, but it would be interesting to find out and not overly onerous to perform.