Duplicate events caused by large adjustments to event times during reassignment

[tjd2002]

@hrjoo and I discovered that we only saw duplicate events on channels there had been a reassignment of events (to that channel) with a large delta_t (typically -18 samples, larger than the detect_interval of 10). We suspect that such large time adjustments are incorrect. Probably there needs to be some dependency on detect_interval in or near this line:

ml_ms4alg/ml_ms4alg/ms4alg.py

Line 429 in 65f8624

dt=tc_peak_times[assigned_channel_within_neighborhood][k]-tc_peak_times[m_central_rel][k]

NB: we still see duplicate events in ml_ms4alg v.0.2.1 (with the per-neighborhood duplicate event deletion provided by 0a8e350). This is probably because that fix does not address duplication across neighborhoods.