On the theory of the surface-stress induced entrainment at a buoyancy interface (toward interpretation of KP and KPA experiments)