Fix angle generation not working for StackedAreaDefinitions

[djhoese]

The optional caching functionality of the angle generation does not work for geometry types that can't be serialized. This logic was able to serialize StackedAreaDefinitions but was only looking for AreaDefinitions. This PR makes it so StackedAreaDefinitions are also cache-able.

  File "/home/a001673/usr/src/satpy/satpy/modifiers/angles.py", line 220, in _hash_args
    arg_hash.update(json.dumps(tuple(hashable_args)).encode('utf8'))
  File "/home/a001673/mambaforge/envs/satpy-3.10/lib/python3.10/json/__init__.py", line 231, in dumps
    return _default_encoder.encode(obj)
  File "/home/a001673/mambaforge/envs/satpy-3.10/lib/python3.10/json/encoder.py", line 199, in encode
    chunks = self.iterencode(o, _one_shot=True)
  File "/home/a001673/mambaforge/envs/satpy-3.10/lib/python3.10/json/encoder.py", line 257, in iterencode
    return _iterencode(o, 0)
  File "/home/a001673/mambaforge/envs/satpy-3.10/lib/python3.10/json/encoder.py", line 179, in default
    raise TypeError(f'Object of type {o.__class__.__name__} '
TypeError: Object of type StackedAreaDefinition is not JSON serializable

• Tests added
• Fully documented

[djhoese]

@mraspaud I could/should update the configuration documentation about how SwathDefinitions and StackedAreaDefinitions are not cache-able, but maybe StackedAreaDefinition could be updated in pyresample to combine the hashes of its sub-areas to generate its own hash?

[djhoese]

Oh maybe that is already done. Let me check.

[djhoese]

Done. Fixed it so StackedAreaDefinitions can be hashed and cached in the angle generation.

[codecov]

Codecov Report

Merging #1946 (482b9e3) into main (2067358) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main    #1946   +/-   ##
=======================================
  Coverage   93.53%   93.53%           
=======================================
  Files         279      279           
  Lines       41276    41317   +41     
=======================================
+ Hits        38606    38647   +41     
  Misses       2670     2670           

Flag	Coverage Δ
behaviourtests	4.77% <6.09%> (+<0.01%)	⬆️
unittests	94.07% <100.00%> (+<0.01%)	⬆️

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

Impacted Files	Coverage Δ
satpy/modifiers/angles.py	96.68% <100.00%> (+0.03%)	⬆️
satpy/tests/test_modifiers.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2067358...482b9e3. Read the comment docs.

[mraspaud]

LGTM, just a minor styling comment. Thanks for taking the time to clean up a bit also!

Another think (that doesn't need to be adressed in this PR): could _hash_args be replaces with dask's tokenize? otherwise I'm wondering if a try/except would be more future-proof than the series of isinstance checks.

[djhoese]

Regarding tokenize, I think it could be, but we may have to look into how it handles object types that it doesn't understand. Meaning, does it do hash(obj) or does it just not hash them or something else. I think we can register new types with the tokenize function in dask if we have to.

Regarding a try/except, that would be nice but I'm not sure how to do it cleanly because we are optimistically running hash(obj) if a geometry object otherwise we just tell json to serialize the object which could be containers. I can't put a hash(obj) in a try/except because it would succeed for most objects.

[coveralls]

Coverage increased (+0.006%) to 93.999% when pulling 482b9e3 on djhoese:bugfix-angles-stackedarea into 2067358 on pytroll:main.