ROI tests

[tcompa]

After closing #23 and #22, we should add some tests (especially for the functions in lib_regions_of_interest.py), and fix the illumination-correction unit test (currently marked as xfail in pytest).

[tcompa]

Basic tests of functions in lib_regions_of_interest are in-place. The relevant test (below) is iterated for level=0,1,2 and coarsening_xy=2,3,7 (where 7 is an extreme case).
https://github.com/fractal-analytics-platform/fractal/blob/a1f86060bcc897ca3a99c78937f2633d5a8550ae/tests/test_unit_ROI_indices.py#L78-L130

In the output indices, we only check some basic properties, like:

• Verify that there are no "holes" (pixels which are not part of any ROI);
• Verify that ROIs' shapes only differ by a number of pixels <coarsening_xy.

The full output is

Original shape: (4, 4320, 5120)
coarsening_xy=2, level=0
Expected shape: (4, 4320, 5120)
FOV-ROI indices:
[0, 4, 0, 2160, 0, 2560]
[0, 4, 0, 2160, 2560, 5120]
[0, 4, 2160, 4320, 0, 2560]
[0, 4, 2160, 4320, 2560, 5120]

.
Original shape: (4, 4320, 2560)
coarsening_xy=2, level=1
Expected shape: (4, 2160, 2560)
FOV-ROI indices:
[0, 4, 0, 1080, 0, 1280]
[0, 4, 0, 1080, 1280, 2560]
[0, 4, 1080, 2160, 0, 1280]
[0, 4, 1080, 2160, 1280, 2560]

.
Original shape: (4, 4320, 1280)
coarsening_xy=2, level=2
Expected shape: (4, 1080, 1280)
FOV-ROI indices:
[0, 4, 0, 540, 0, 640]
[0, 4, 0, 540, 640, 1280]
[0, 4, 540, 1080, 0, 640]
[0, 4, 540, 1080, 640, 1280]

.
Original shape: (4, 4320, 640)
coarsening_xy=2, level=3
Expected shape: (4, 540, 640)
FOV-ROI indices:
[0, 4, 0, 270, 0, 320]
[0, 4, 0, 270, 320, 640]
[0, 4, 270, 540, 0, 320]
[0, 4, 270, 540, 320, 640]

.
Original shape: (4, 4320, 5120)
coarsening_xy=3, level=0
Expected shape: (4, 4320, 5120)
FOV-ROI indices:
[0, 4, 0, 2160, 0, 2560]
[0, 4, 0, 2160, 2560, 5120]
[0, 4, 2160, 4320, 0, 2560]
[0, 4, 2160, 4320, 2560, 5120]

.
Original shape: (4, 4320, 1706)
coarsening_xy=3, level=1
Expected shape: (4, 1440, 1706)
FOV-ROI indices:
[0, 4, 0, 719, 0, 853]
[0, 4, 0, 719, 853, 1706]
[0, 4, 719, 1439, 0, 853]
[0, 4, 719, 1439, 853, 1706]

.
Original shape: (4, 4320, 568)
coarsening_xy=3, level=2
Expected shape: (4, 480, 568)
FOV-ROI indices:
[0, 4, 0, 239, 0, 284]
[0, 4, 0, 239, 284, 568]
[0, 4, 239, 479, 0, 284]
[0, 4, 239, 479, 284, 568]

.
Original shape: (4, 4320, 189)
coarsening_xy=3, level=3
Expected shape: (4, 160, 189)
FOV-ROI indices:
[0, 4, 0, 80, 0, 94]
[0, 4, 0, 80, 94, 189]
[0, 4, 80, 160, 0, 94]
[0, 4, 80, 160, 94, 189]

.
Original shape: (4, 4320, 5120)
coarsening_xy=7, level=0
Expected shape: (4, 4320, 5120)
FOV-ROI indices:
[0, 4, 0, 2160, 0, 2560]
[0, 4, 0, 2160, 2560, 5120]
[0, 4, 2160, 4320, 0, 2560]
[0, 4, 2160, 4320, 2560, 5120]

.
Original shape: (4, 4320, 731)
coarsening_xy=7, level=1
Expected shape: (4, 617, 731)
FOV-ROI indices:
[0, 4, 0, 308, 0, 365]
[0, 4, 0, 308, 365, 731]
[0, 4, 308, 617, 0, 365]
[0, 4, 308, 617, 365, 731]

.
Original shape: (4, 4320, 104)
coarsening_xy=7, level=2
Expected shape: (4, 88, 104)
FOV-ROI indices:
[0, 4, 0, 44, 0, 52]
[0, 4, 0, 44, 52, 104]
[0, 4, 44, 88, 0, 52]
[0, 4, 44, 88, 52, 104]

[tcompa]

I'm closing this now, but feel free to re-open to suggest other (more thorough) checks on the FOV-ROI indices.