The scrambler now generates optimal move sequences. 96.5% of optimal scrambles are 18 moves or shorter, and none require more than 20 moves. Less scrambling, more solving. As always, cube positions are chosen with equal probability.
Because of its compute-intensive nature, it is necessary to rate-limit the optimal scrambler. During times of heavy traffic it will fall back on a near-optimal scrambler to keep up with demand. These will be identified by the tag [Suboptimal].
Made the scrambler more suitable for timed practice by removing visualization.