If we assume that the blade is initially undeformed, and if the purpose of the test is to detect deformation caused by improper mounting, then I'm not sure all this is needed. But perhaps I'm missing something here.
Let's first take the simplest scenario: a blade with flat, parallel sides (such as Wilson Coronation Ace). We first test the unmounted blade for deformation by laying a straight edge against each side of the runner (first one side, then the other). Assume all is good: the straight edge will lie flat against the runner; there will be no visible gaps between the straight edge and the runner; and the straight edge will not rock against the runner.
We mount the good blade.
We then test the mounted blade again by laying a straight edge against each side of the runner (first one side, then the other). If the mount is good, then the straight edge will lie flat against the runner; there will be no visible gaps between the straight edge and the runner; and the straight edge will not rock against the runner. If the mount is not good, then the blade can be deformed: bowed along one axis, twisted along another axis, or both. If so, then the straight edge will not lie flat against the runner; and depending on the scenario, there will be visible gaps between the straight edge and the runner, or the straight edge will rock against the runner, or both.
With a more complicated blade geometry, for example a blade with concave side honing (such as Wilson Gold Seal), as you have noted, we are then dependent on the stanchions for reference surfaces (which, as you have noted, may or may not in fact be good reference surfaces). Assuming they are, then we should be able to place a straight edge across the three stanchion surfaces to check for deformation similar to the above procedure. We would repeat these tests, checking at multiple locations across the stanchion surfaces.
Is the point of your test setup (and the PBHE test setup) to make it easier to detect twist because you effectively magnify the difference between measurement surfaces? What about bowing along a different axis? That is, the blade is (exaggerated)
pick
(
heel
The PBHE tester doesn't pick up bowing either.
Hope this is clear. If not, I'll make some drawings.