Windtunnels just like flow dynamic computer codes are alwys just a model of reality, and none gives exact results when compared to free flight measurement. Many phenomena have been discovered in free flight and could not be confirmed in the wind tunnel or by calculation. It is hard to produce a real free air stream in a wind tunnel, there always is some turbulence created by the fan or by the walls of the air intake, there is always some slight speed gradient close to the walls, there is always some restriction in the air flow if you add walls. Vice versa any computer code can just predict what has been part of the model from where formulas have been derived.
A recent example is that a turbulent boundary layer has been found to become laminar again under certain conditions, while mathematics tells you that this is impossible, and wind tunnels can not reproduce this.
Any theory is only good until a better one comes along. Any wind tunnel is only as good as the criteria it was designed for. Any measurement method (pressure by piezo sensors, speed by LDA, drag by wake analysis...) only works if you exactly know what phenomena to measure. There is not a lot of static flow in a steady flight, there is a lot of small amplitude high frequency variations in the parameters required to fully describe certain phenomena.
It looks like we understand aerodynamics for some hundred years, but in fact we are still learning something new every year.