The cooling masters guys and in my test, I'm pulling with the fans, so it can't be heat dump from the fans. Also in the coolingmasters test, they show this affect occurring to a very minor degree with the nexus fans at 12V, it's not really flat past 1GPM until they tested with the Panaflos at 7V.

And we're not talking huge differences, but something and perhaps more with certain radiators than others.

I just see this as another possible optimization. The radiator manufacturers could either build in their own bypasses to emphasize performance in one area, or the user could do it with plumbing/valves.

I'm certainly seeing this in my closed loop testing now and I'm repeating my tests, so it's checking out.