omit overlapping shoeboxes because we can't handle them yet

[dwpaley]

No description provided.

[nksauter]

As I understand it, this code removes dials spotfinder spots when the combined spot+three-pixel border overlaps with another. I would rather not throw out spotfinder spots, for reasons already discussed--it seems like an arbitrary decision, not easily explained in a paper. My preference would be to modify the code that creates the three-pixel border, just to assure that two refls do not claim the same border pixels. Perhaps it could be first come, first served. You could use a dictionary to keep a list of all total pixels claimed by previous spots in the list. I would rather go with this alternate idea.

[nksauter]

Alternate approach suggested above.

[dermen]

I do this here - the first ROI in the list of ROIS gets preference and can claim all of its pixels.. then if a subsequent ROI tries to claim any already claimed pixels, the ROI is shrunk by a pixel in each direction until it is no longer asking for claimed pixels .

https://github.com/cctbx/cctbx_project/blob/6ec995a7c3565e2b077b6e5d043f6cced8952699/simtbx/diffBragg/utils.py#L657