participants in random condition may fail to get regimes for breathing

[noah10]

At https://github.com/EmotionCognitionLab/pvs/blob/ebf85cbf2bde9264bd574d0214f10c82467627d6/client/src/regimes.js#L230, we're in a rare state where there are no regimes whose 90% CI bounds overlap the target average coherence. In this case we're just supposed to pick the regime whose average coherence is closest to the target average coherence. Unfortunately, we try to do that using the wrong key (avg_coherence, rather than mean), making our attempt futile and generating an error.

[choipaul92]

Tested and Breathing Practice Worked after Deleting old Heartbeam and Reinstalling new Heartbeam!