Fix bbox creation for SwathDefinitions with NaNs

[djhoese]

I ran into this with some VIIRS data and some AVHRR data. In one case I had a couple rows/scans of data that had NaNs in them so some of the bbox coordinates were NaN. This resulted in bbox coordinates being returned, but being invalid (had NaNs). In another case the top couple scans of the swath were all NaNs so you'd get some hard to decipher error about NaN values. Sometimes these only made themselves obvious when you made a Boundary/Polygon and tried to do some calculation with them.

This PR fixes these cases or improves the error messages by removing any bbox points that have NaN for one of the coordinates.

CC @ghiggi

• Tests added
• Tests passed
• Passes git diff origin/main **/*py | flake8 --diff

[codecov]

Codecov Report

Merging #479 (ac45d86) into main (80aeb8e) will increase coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #479      +/-   ##
==========================================
+ Coverage   94.31%   94.32%   +0.01%     
==========================================
  Files          74       74              
  Lines       12855    12890      +35     
==========================================
+ Hits        12124    12159      +35     
  Misses        731      731              

Flag	Coverage Δ
unittests	94.32% <100.00%> (+0.01%)	⬆️

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

Impacted Files	Coverage Δ
pyresample/geometry.py	87.50% <100.00%> (+0.10%)	⬆️
pyresample/test/test_geometry.py	99.54% <100.00%> (+<0.01%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[coveralls]

Coverage increased (+0.02%) to 93.851% when pulling ac45d86 on djhoese:bugfix-swath-bbox-nans into 80aeb8e on pytroll:main.

[pnuu]

LGTM