We assume the same thing as what DaveAtkins states. Here's why:
The exterior flange of the wall girt is obviously braced laterally by the exterior skin. For suction, the rods hanging down from above have good tension support, but questionable compressive stiffness. In addition to that, the rods are usually bolted through the web, and not connected to the interior flange.
But if you assume that the purlin has just a bit of downward sag in it, then any suction on the wall would immediately initiate a compressive force in the interior flange which would try to buckle downward. The sag rod through the web, and the skin attachment to the exterior flange create a sort of couple that would resist rotation.
And resisting rotation is almost as good at bracing a beam as resisting lateral translation of the compression flange (according to Yura).