I agree that the fighter pilots would have driven their kites closer to the limits than tought at flight school. As you indicate yourself by your phrase, aerodynamic calculations are never as accurate as to predict reproducable stall speeds. But this may go both ways that is that calculations are either pessimistic (stall would occur later than calculated) or also optimistic (stall would occur sooner than calculated). So it may be the way you stated it (the pilots drove their plane closer to the limits than what calculations would have predicted) but it may also be the other way around.

Principally I would guess that stall speeds taught to the cadets were obtained experimentally. And keep in mind that pilots appreciated when they got a feedback from the plane (e.g. buffeting) when they got close to the stall limit.