Hmm...if you look at a ski racers trajectory in two components, down the hill and across the hill, it becomes obvious what's going on. A ski racer wants to maximize the down the hill velocity to get to the finish line in as short a time as possible. But he/she has to get around the gates so across the hill the trajectory is accelerating side to side. After going around a gate the skier has to slow his/her side cross slope velocity to zero then accelerate the other direction.
A stivot in the hands of the pro skier maximizes the lateral deceleration while minimizing forward deceleration. When the racer reaches the gate the skis are aligned with the velocity vector and the edges set just like you would do at the top of a turn and the skis start to carve resulting in rapid acceleration across the slope the other way.
The key here is the skis have to be aligned with the direction of travel across the snow, or at least close to it, before setting the edge. If not then the resulting skid will bleed off some of the velocity down the hill and the skier will lose time. Watch the videos very carefully and you will see the while in the stivot the skis are pointed either down the fall line or slightly toward the next gate. Skidding with skis in this orientation does not bleed off downhill speed and can even result in some downhill acceleration.
Just an engineer's perspective.
Step back a year and compare the speeds of the carve and the spivot. Carve was faster. This indicates that the spivot scrubs speed each and every turn. Today with the course set to disadvantage carving spivot is faster. This only indicates that a spivot works very well as a compromise to make the turn when it can't be carved.