Fixing automatic dot counting code:
z-sections:
small number of sections gives correct assignment. Dots that extend into upper layers (line extends to the left) are not tracked, as they are marked when they first entered those layers. Dots that extend into lower layers are tracked the first time the appear. Dots only seen in one channel have been removed. Dots have that have multiple maxima have been split into separate dots.
Z-section connect gets confused and keeps all dots: error results from uint16 not being large enough to index all dots.
This is now fixed.
Also added numerous clear statements inside main Unsupervised_dotfinding and supporting function calls. This keeps the code under the memory maximum and can run through a slide’s worth of data without crashing.