The only person really properly testing watercooling gear is BillA. I think he doesn't do it for a living, however
so the whole community can't really rely on only him. The problem is that end user numbers are really useless, and even the numbers I can generate at the moment are not that much better. I am working on some of my test setup's flaws, however. Roundups from enthusiast sites become far more useful if flow data and multiple tests are conducted. I pretty much know if I get a good paste application and a good mount on any of the blocks I regularly use; I wouldn't know this if I were mounting it for the first time. I trust Joe (on this site) because he is up front about the limitations of his testbed and says straight up that the testing isn't an exact science. In my mind that suggests that he is also doing his best to limit variability when he can. Reviewers that don't realize their limitations are also less likely to minimize other errors IMO. This may just be bias from working in a chemistry lab though.
I think that physical properties, pressure drop, and airflow vs. static pressure numbers for your radiators would go a LONG way even without any real performance data. The same is true for knowing the properties and dimensions of your block. I recognize the bias of manufacturers, but it would be nice to at least show that some testing and QC and R&D go into the final product. There is charm to the DIY aspect of water cooling, but it is also nice to know that you are buying something that is operating at optimum performance (either by trial and error or by some engineering know how) .