Fix bug that would add spurious nan lines in gradient search

[mraspaud]

This PR fixes #642.

The symptom was that every other line, or all lines, of chunks that were on the edges of a geos disc would be nans. That was because the cursor in the gradient search algorithm was not moved along x when a pixel was skipped because of invalid target coordinates.

• Closes Gradient search skips some data when resampling from geos to mercator #642
• Tests added
• Tests passed
• Fully documented

[mraspaud]

Fixed images from the use case in #642:

[pnuu]

I restarted CI, there was a hickup with Conda.

[pnuu]

LGTM.

Simple fix, but must've been a pain to hunting the bug 😅

[djhoese]

Looks good from what I understand. I had one question just to clear something up for me, but otherwise good to merge.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 93.96%. Comparing base (d33686e) to head (7ac934f).
Report is 4 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #643   +/-   ##
=======================================
  Coverage   93.95%   93.96%           
=======================================
  Files          86       86           
  Lines       13541    13546    +5     
=======================================
+ Hits        12723    12728    +5     
  Misses        818      818           

Flag	Coverage Δ
unittests	93.96% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[coveralls]

coverage: 93.656% (+0.002%) from 93.654%
when pulling 7ac934f on mraspaud:fix-gradient-geos
into c499af0 on pytroll:main.