Fix class thresholding

the threshold of 100 transcripts to call a cell positive for Slc17a7 is probably too high. either Slc17a7 spots need to be thresholded for intensity, as was done in prior versions, or we should increase the threshold Currently cells that are high for Gad2 and Slc17a7, of which there are many, are labeled as unassigned even though most of them are obviously inhibitory. The definition of inhibitory should be high for any subclass gene or Gad2, regardless of the Slc17a7 expression.

Could also consider using GMM for this