Typical bicycle industry stupidity. Useless drivel that would be skewered by legitimate engineers.
The pedaling rider on the bicycles is a gargantuan confounding factor... especially when you’re talking about a one second difference over the course of a km. If you think his movement would be identical on both bicycles, you’re nuts. Even if the drive trains were identical, you’d be nuts. (Granted, to their credit, they did acknowledge that the drivetrains were different).
The proper way to study this would be to have a static object on the frames representing the rider (eg a rider or a dummy), and blow wind at these things at increasing velocities (up to the maximum speed a rider may propel a bicycle) and measure the force of the drag. This should be done repeatedly, with the rider in different positions each time (eg pedals mid-stroke or at 3 and 9 o’clock, hands in hoods, hands in drops, etc). Drag forces should be measured for each velocity and each rider position. But having some bald-headed twirp get on a bike and pedal into the wind is a surefire way to introduce error and skew results.