Gerwin,
It is possible to use a overly large pump that puts so much heat in the water that it overwhelms the rads ability to remove it to the point that the higher flows still result in higher temps.
But on the other hand, .53-1.2lpm are the other extreme as they are very low flows. That is clearly why he had temps of 48-49 C. Those are not real good air temps, and with water they are sad. The graph, if anything, proves the value of higher flows. If he was the carefull tester you say, why didn't he test at more and higher flow points? Had he also tested at say 2, 3 & 4lpm the benifit of higher flow rates (within sane levels) would have become very clear. And 2-4lpm flow is still modest. But 4lpm would have gotten his temps out of the high fortys, and down somewhere in the 30's if his rad was of any value at all.
If you go to this link you will find a testing article by Bill Adams on the proper testing of waterblocks (amoung others). His testing maintains very tight controls. Yet his conclusions on how high a flow rate is helpfull are very differant than the ones that graph would have you believe. Here is the link.
www.overclockers.com
I'm not looking for endless debate here. Just thought I'd offer a link showing the other side of the coin. And by a man who isn't a fan of extreme pump set ups. I do wish I could read german as I've seen enough referances and just pix at their sites to know there are some good things going on there.