That's just one aspect, one which you'll need to take into account once you're reaching a certain level of wind resistance.
I would not be surprised to learn that the difference in wind resistance as seen by say, a 23mm compared to the 27mm version of the same tyre is far bigger than 10 to 15 Watt at higher speeds in a wind free environment already.
Either way, let's just assume that windresistance was not the target of the test.
If you want to measure rolling resistance than the protocol should be the same for every tyre within the apples to apples and oranges to oranges range. Carrying out such tests is far more easy to do and far more precise in a controlled environment.
From that prespective the lab tests as carried out by Tour magazine are scientifically far more valid than any such test carried out on a road.
Whether or not the tyre sees a fixed load (brick of lead fror instance) or a rider + a road as load isn't relevant at all for measuring rolling resistance either for as long as the load is the same for all devices under test. Something you can't be sure of when testing outdoors on a strip of road.
Point is that the relative order of difference between the test participants should and will be absolutely the same in both tests. The absolute figures may differ yet the differences should remain.
In short, there are a number of remarks and conclusions in that article that just goes to show that there's nothing new under the sun; setting up tests is one thing, drawing the right conclusions is quite another....
Plus ca change.....