Best way to tabulate data when sync_events and bmi3d_events are different

When `len(exp_data['sync_events']) != len(exp_data['bmi3d_events']`, 'bmi3d_events' is more trustworthy.

But `tabulate_behavior_data()` uses 'sync_events' (i.e. 'events') to segment trials. Not sure what causes this bug in 'sync_events' - maybe related to pause bug or non-zero values in `exp_data['sync_events']['data']`?

To avoid the bug, `get_ref_dis_frequencies()` uses 'bmi3d_events' but then the number of trial segments doesn't match up with the df from `tabulate_behavior_data()`. Potential fixes: handle data differently when # of 'events' and 'bmi3d_events' doesn't match, or find another method for getting the frequencies that works either way