Refactor resampler module to submodules

[pnuu]

This PR refactors the satpy/resample.py module to a satpy.resample subpackage and splits the contents to several modules.

At least the warnings in the tests should be fixed, but this is the first version to see if there are other things to consider. Also some of the imports could be moved to where they are used. And the docstring should be fixed with the new import paths.

Triggered by #3123

• Closes #xxxx
• Tests added
• Fully documented
• Add your name to AUTHORS.md if not there already

[djhoese]

The other thing about this PR that I'm realizing is that some of the resamplers (ewa maybe? kdtree nearest maybe?) exist in pyresample in a way that should be useable. Or at least that was my intention before I stopped working on pyresample 2.0. Maybe it never got all the way there (as we already know I'm wrong about the caching functionality implemented in pyresample).

[codecov]

Codecov Report

Attention: Patch coverage is 92.11747% with 51 lines in your changes missing coverage. Please review.

Project coverage is 96.25%. Comparing base (55328e2) to head (80296ee).
Report is 38 commits behind head on main.

Files with missing lines	Patch %	Lines
satpy/resample/kdtree.py	87.65%	20 Missing ⚠️
satpy/resample/base.py	89.36%	10 Missing ⚠️
satpy/resample/native.py	91.96%	9 Missing ⚠️
satpy/resample/bucket.py	94.94%	5 Missing ⚠️
satpy/area.py	75.00%	3 Missing ⚠️
satpy/coords.py	94.00%	3 Missing ⚠️
satpy/resample/__init__.py	88.88%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##             main    #3124    +/-   ##
========================================
  Coverage   96.24%   96.25%            
========================================
  Files         421      429     +8     
  Lines       57671    57773   +102     
========================================
+ Hits        55506    55608   +102     
  Misses       2165     2165            

Flag	Coverage Δ
behaviourtests	3.80% <2.16%> (-0.12%)	⬇️
unittests	96.34% <92.11%> (+<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.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[coveralls]

Pull Request Test Coverage Report for Build 15348574923

Details

• 601 of 652 (92.18%) changed or added relevant lines in 30 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.007%) to 96.356%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
satpy/resample/init.py	9	10	90.0%
satpy/area.py	10	13	76.92%
satpy/coords.py	48	51	94.12%
satpy/resample/bucket.py	94	99	94.95%
satpy/resample/native.py	103	112	91.96%
satpy/resample/base.py	85	95	89.47%
satpy/resample/kdtree.py	142	162	87.65%

Totals
Change from base Build 15307805052:	0.007%
Covered Lines:	55895
Relevant Lines:	58009

---

💛 - Coveralls

[pnuu]

I think this is now ready. Unless we come up with a better name for the base module.

[djhoese]

One more comment for making things private, but also what do you think about moving get_area_file/def functions and add_xy_coords and add_crs_xy_coords to a utils.py module? Or even go as far as the get_area_* functions to one module outside of the resample subpackage (since they are used by readers and users) like satpy.area or satpy.area_utils or satpy.area_config? And add_*_coords to another module inside or outside of resampling? Like satpy.coords or satpy.coords_utils?

[djhoese]

Oh another thought: if we imagine that the base resampler class(es) will all end up moving to pyresample and possibly even the "prepare_resampler" or "get_resampler"-style functions would be moved. Does that change the naming? Or rather, maybe those types of functions should not be in base.py? If the base class went away then these helper/utility functions would be in a base.py when they don't do anything "base"-like.

[pnuu]

One more comment for making things private, but also what do you think about moving get_area_file/def functions and add_xy_coords and add_crs_xy_coords to a utils.py module? Or even go as far as the get_area_* functions to one module outside of the resample subpackage (since they are used by readers and users) like satpy.area or satpy.area_utils or satpy.area_config? And add_*_coords to another module inside or outside of resampling? Like satpy.coords or satpy.coords_utils?

I'll have a look at this today evening after I reach the hotel in Milan, where I need to stay overnight due to a cancelled flight.

[djhoese]

Awesome. I noticed one small thing that could be changed and cleaned up. The only other discussion we should maybe have is how many satpy/X_utils.py modules do we have before making things a series of utils/X.py modules? Or would it be better to drop the _utils.py part of the name and just call them satpy/area.py and satpy/coords.py?

[djhoese]

Awesome! Just. Awesome.

I think the most obvious win from this PR is how many things are loading from the area and coords utils instead of the resample module/subpackage.

I made a couple inline comments about long term design decisions. The only other remaining question for me is should satpy/area_utils.py and coords_utils.py have the _utils removed?

[pnuu]

The only other remaining question for me is should satpy/area_utils.py and coords_utils.py have the _utils removed?

These were renamed in 5bbd4f1 and 0b2be45

[djhoese]

Looks good. I think the only thing I want to wait for is other people's feedback. Particularly on the satpy/area.py and satpy/coords.py naming (previously with the _utils.py suffix). If we're being pedantic I think there is an argument for plural versus singular naming (areas/area versus coords/coord) and consistency.

[mraspaud]

LGTM, thanks a lot for this huge work. Always nice to see refactoring monolith code in separate modules