Reinstating nuclei detection and nuclear masking

[vreuter]

Spot use quality goes way up if we filter for inclusion in nuclei

Related:

• [ ] #55
• [x] #34
• [x] #32
• [x] #30
• [x] #243
• [ ] #244
• [x] #245
• [x] Ensure unique nucleus identifier for each mask -- @TLSteinacker
• [x] Need nuc_slice value (or 3D) to influence how we take the reference image for coarse drift correction
• [ ] #250
• [x] visualisation of spot count before and after nuclei-overlap filtration
• [ ] ensure that thresholding (NucDetector.segment_nuclei_threshold) is well-behaved and passes tests also (like for Cellpose)
• [ ] #252