The test would be done with the TG locked up. (BTW, an easier way to lock up is to use something like
this to pull the heel down on the stock before you close the trigger guard.) The basic idea is to prevent the receiver from rotating sideways in the stock, even a little bit. Unless it's rear or double lugged, the only parts of the receiver that are in contact with the
inside surface of the stock are the outer faces of the receiver legs and the section just forward of the legs. If the fit there isn't extremely tight, a small amount of rotation is possible (which is what the test of movement of the heel over the stock is designed to disclose). And if there's a little rotation at the receiver, imagine how much displacement there is way down at the end of the stock/barrel contact point (front band lip/ferrule). The tension intended to guide the barrel back to it's rest point on the ferrule would be guiding it to a different place every time, not a good thing.
Even if the heel is really tight against the top of the stock, if there's some slop between the legs or the section forward and the stock, you can still get a bit of sideways movement there and a lot at the ferrule.
Using shims to raise the heel even further off the stock would make locking up even harder or maybe impossible. But using the "L" shaped shims just inside the stock would tighten the fit without tightening the lockup. In any case, thinking of that tension as part of the "see-saw" motion has the ferrule end of the stock being pulled down (creating tension) as the back end of the stock is pulled up to meet the heel. Since that's what's happening to you, I bet you've got the tension, just not the clearance so it can be demonstrated or operational.