Thank's Mr701. Excellent comments, in the video, that explain why it happened.
This close to the ground and this slow, we don't want to be looking down at the ball. Simply correcting with rudder, static reactive, to keep the longitudinal axis pointed down the runway may correct in time. Or maybe not.
Notice that the rudder is not wiggling dynamically and proactively in the rear view. Dynamic proactive rudder movement to prevent the misalignment or to prevent the roll to begin with is more effective.
He is hovering out of low ground effect; very tricky and power dependent. If we had a video of the empty copilot antitorque pedals of a helicopter in a hover, we would see continuous left, right, left, right antitorque pedals movement.
Normally powered light airplanes can certainly get down short final without wagging the tail a bit. Because they are faster, they may catch the upset with aileron or better with rudder only. The best way to stay ahead of the airplane and prevent upset, however, is to continuously move rudder only a bit to bracket out any problem from any source. It is amazing how the brain causes a bit longer or shorter a wiggle depending on need. I hate to say, "Use the force, Luke Skywalker," but that is exactly what is correlating between brain and hand even though we just see continuous movement.
If we wait for the airplane or the upset or whatever, we will have to react to correct. That in no way can be as fast and effective as preventing the problem continuously with dynamic proactive rudder movement any time we are near the ground or on the surface.
Download my free "https://tinyurl.com/Safe-Maneuvering" e-book.