Nwb integration

[mhturner]

Adds the option of saving .nwb file formats instead of the old visprotocol-based .hdf5 files. A major change in organization is that now each epoch run is associated with a separate .nwb file, instead of a single .hdf5 file holding all of the subjects and epoch runs for that day's experiment. This is required by .nwb format specifications. This necessitated big changes in the GUI that I couldn't figure out how to smoothly switch between, so there are two different GUI versions.

The way to switch to .nwb is to use the NWBData class in data_nwb, e.g. in your lab-pack data.py module you can add:

class Data(data_nwb.NWBData):
    def __init__(self, cfg):
        super().__init__(cfg)  # call the parent class init method

assuming your config file looks like this:

module_paths:  # relative to the labpack directory specified in stimpack.experiment's path_to_labpack.txt
  data: labpack/data.py  # module for data class. Class name should be "Data"

Then run the nwb-specific GUI with:
python gui_nwb.py

I tested these two ways of running stimpack and both seem to be working OK, but I would recommend testing thoroughly before deploying in your own labs, where people are still using the old file format. We have switched exclusively to .nwb in my lab.