2D classification of template-picked particles causing issues

The initial 2D classification did not use a circular mask, so I did not use one in the second iteration of 2D classes.

I went ahead and tried using a circular mask. The classes look better than J112, but still much worse than J106.