Thanks for taking the time to do and post, this is fascinating!

Regardless of the results, I will continue to use rack&pans instead of backpack for reasons of sweat and comfort, but this is helpful.

It reminds me of something I read in Bicycling magazine how after Lemond kicked ass in TdF with aerobars, people did coast-down tests with aero- vs drop-bars, and it was in that case a clear difference. In this case, the results seem to make sense with the idea that the backpack is in the aero 'shadow' so it doesn't have an effect.

If that is the case, perhaps the tiny boost in speed you're seeing is due to extra mass. As even Galileo knew, bowling balls do drop faster than golf balls, because their greater momentum is less affected by air resistance. And as we all know, nobody can descend like a fat man! If you replaced the cargo with actual stuff, I bet you'd see another small speed boost.

I think I need to design and patent a system that has robotic arms to reach out to the side of the road and grab rocks for ballast and stuff them into my panniers whenever a bubble-level shows I am cresting a hill; then when I get to the bottom, the robot arms can pull all the rocks out so I'll be ready for the next uphill.