bug from hell when running HMMs on jetstream

[danielsvedberg]

Given a specific dataset, phmm.query_units() will return n**+1** units only when the area='GC' argument is supplied. This causes

blechpy/blechpy/analysis/poissonHMM.py

Line 513 in 8a5ab8b

def get_hmm_spike_data(rec_dir, unit_type, channel, time_start=None,

to throw an error at line 554. Only when I run fit_hmm_mp on a specific dataset on jetstream. Other datasets don't seem to have problems.

I have systematically excluded the following possible causes:

• different # of units in HDF5 and unit table in dat. not that it should matter in this case
• package and/or python versioning, code runs fine on my machine, even in a fresh install of blechpy in a fresh venv
• syntax error: given that the area='GC' argument causes this to fail, lines phmm lines 629-636 should be responsible for this bug. despite this, I can't find anything wrong