[WIP] add UnitSeries neurodata_type

[ajtritt]

This is a proposal/work-in-progress for addressing #194. It adds a new type, called UnitSeries, that would replace Clustering. It extends TimeSeries and requires that TimeSeries.data be an int. It also adds an additional dataset, called units that is a DynamicTableRegion that points to Units table.

[tjd2002]

Would this allow two or more spikes from different units with the same timestamp? (This can happen in real data)

[ajtritt]

@tjd2002

Would this allow two or more spikes from different units with the same timestamp? (This can happen in real data)

Yes, there are no restrictions on unique time stamps in TimeSeries

[bendichter]

I'm fine with the name change, and I like that this now inherits from TimeSeries, but it would be nice if there was some kind of link to SpikeEventSeries (#239). As it stands now, if you had multiple UnitSeries and multiple SpikeEventSeries objects (e.g., 6, one for each of 6 shanks), it is unclear how you would indicate which SpikeEventSeries object is connected to which UnitSeries object. I guess you could do it implicitly if both had a field for referencing the regions of the Units table, and you implied that objects that referenced the same Units table region were connected.