[000939]: TimeIntervals tags written incorrectly

[stephprince]

This dandiset issue came up in NeurodataWithoutBorders/pynwb#1990, where it appears that there may have been an issue during data conversion with the tags column of the "nrem" intervals table.

It seems the tags were incorrectly parsed into separate strings when the data was written (e.g. the tags go from [['8'], ['9'], ['1', '0'], ['1', '1']] instead of ['8', '9', '10', '11']).

I will contact the data contributors to determine if the dandiset can be updated.