"What oil(s) is being compared against RP in all of those tests? Another synthetic oil? Or a conventional oil? "
We don't know what oil is being used (in many cases), nor the weight of that oil, and more importantly: we don't know how many miles on that oil. It is possible that one is comparing a fresh batch of 0W-30 against a batch of 20W-50 with 10,000 miles on it.
The only FAIR way to do this test is to start with one brand of fresh oil (with zero miles), run the dyno, change brands of oil but keep the weight the same, run the dyno, change back to another fresh batch of the starting oil, run the dyno, and change back to the second oil (again) and run the dyno. {Changing the filter each time}
By looking at all 4 sets of dyno graphs, one can begin to actually know something about the oils and engine and dyno. By averaging the first and third runs and averaging the second and fourth runs, and comparing both averages, you can see if there is any significant actual difference.
All 4 runs need to take place in less than 1 hour or the data becomes suspect (again). This makes it very difficult to change the tranny oil and differential oil at the same time the engine oil gets changed. And also means you might only get 30%-50% of the advertized gain if you only change the engine oil (if that much).
Idealy, one would like to instrument the car under test, and set up a situation where the engine, tranny, and differential oils can be changed while the engine is running (or briefly stopped) so that one can run all of the tests in a short amount of time, at the same engine, tranny, and differential temperatures, and at the same TIRE temperatures. But this makes the test "expensive" to run (and set up to run).