Refactor of organoid linking over multiplexing rounds

[nrepina]

Create separate functions for registering and linking organoid label map images to share same processing code between Fractal and the Prefect implementation of scMultipleX.

[nrepina]

@jluethi Please take a look at this refactor for organoid linking across multiplexing rounds and let us know what you think and if we missed anything. From our side this refactor is ready to merge, but if there are any further changes from the Fractal side we are happy to add changes with another PR.

The functions that perform registration and linking are now in the new OrganoidLinkingFunctions.py, and these would be the functions useful for the Fractal implementation. I've added descriptions of the functions and parameters, please let me know if there's anything unclear. The calculate_shift, apply_shift, and calculate_matching functions are then called in OrganoidLinking.py to process pairs of well overview images. This file is now primarily dealing with input/output that is specific to FAIM-HCS/Prefect.

A few small notes:

• The apply_shift function can be used for intensity images, not only label images, so the calculated shifts can be applied to the MIP overviews for visualization.
• I found that using a bin = 4 is a good value for subsampling
• I typically use iou_cutoff = 0.6

Thanks, looking forward to your feedback!

[jluethi]

@nrepina Looks great & very informative docstrings! Will work on a Fractal wrapper task for it next week :)