remove 0.8 deprecations from pvsystem.py, modelchain.py, rework temperature model param deprecation

[wholmgren]

• Closes PVSystem.temperature_model_parameters requirement #1030 closes remove deprecated functions in 0.8 #966
• I am familiar with the contributing guidelines
• Tests added
• Updates entries to docs/sphinx/source/api.rst for API changes.
• Adds description and name entries in the appropriate "what's new" file in docs/sphinx/source/whatsnew for all changes. Includes link to the GitHub Issue with :issue:`num` or this Pull Request with :pull:`num`. Includes contributor name and/or GitHub username (link with :ghuser:`user`).
• New code is fully documented. Includes numpydoc compliant docstrings, examples, and comments where necessary.
• Pull request is nearly complete and ready for detailed review.
• Maintainer: Appropriate GitHub Labels and Milestone are assigned to the Pull Request and linked Issue.

I'm still struggling with the right approach for #1030. Here are examples of the behavior in this PR:

In [109]: PVSystem()
/Users/holmgren/git_repos/pvlib-python/pvlib/pvsystem.py:219: pvlibDeprecationWarning: temperature_model_parameters, racking_model, and module_type are not specified. Reverting to deprecated default: SAPM cell temperature model parameters for a glass/glass module in open racking. In v0.9 PVSystem.temperature_model_parameters will be empty dict unless set by temperature_model_parameters or valid combination of racking_model and module_type. Pass non-None value to any of these parameters to silence this warning.
  pvlibDeprecationWarning)
Out[109]:
PVSystem:
  name: None
  surface_tilt: 0
  surface_azimuth: 180
  module: None
  inverter: None
  albedo: 0.25
  racking_model: None
  module_type: None
  temperature_model_parameters: {'a': -3.47, 'b': -0.0594, 'deltaT': 3}

In [110]: PVSystem(racking_model=False)
Out[110]:
PVSystem:
  name: None
  surface_tilt: 0
  surface_azimuth: 180
  module: None
  inverter: None
  albedo: 0.25
  racking_model: False
  module_type: None
  temperature_model_parameters: {}

This change also causes 28 new warnings from just test_pvsystem.py. I'm sure there are more in test_modelchain.py. In practice, I think this only matters to users that call PVSystem.sapm_celltemp, so we could push the deprecation down to this method.

[wholmgren]

Moving the deprecation warning from PVSystem.__init__ to PVSystem.sapm_celltemp works with one caveat: ModelChain.infer_temperature_method cannot infer the SAPM method. A temporary work around could be:

    def infer_temperature_model(self):
        params = set(self.system.temperature_model_parameters.keys())
        # revert this change in 0.9
        if set(['a', 'b', 'deltaT']) <= params or (if not params and self.system.racking_model is None and self.system.module_type is None):
            return self.sapm_temp  # system.sapm_celltemp will assign default keys and emit warning at runtime

Does that all seem reasonable @cwhanse?

[cwhanse]

Suggested change

	'in open racking. In v0.9 '
	'PVSystem.temperature_model_parameters will be empty dict '
	'unless set by temperature_model_parameters or valid '
	'combination of racking_model and module_type. Pass non-None '
	'in open racking. In v0.9, temperature_model_parameters or a valid '
	'combination of racking_model and module_type will be required. '
	' Pass non-None '

I'd prefer to be more specific how to use v0.9.

[wholmgren]

I'm ok with "will be required" if the warning is emitted from PVSystem.sapm_celltemp. I don't like it if the warning is emitted from the PVSystem.__init__ because it suggests a positional argument.

[cwhanse]

PVSystem() has always been a valid, non-warned statement. There's never been a requirement that PVSystem is a complete description of a system according to some model. We can discuss changing that, but it's inconsistent to require temperature parameters and nothing else.

Agree, and point taken. I'm OK with the change proposed for ModelChain.infer_temperature_model.

[wholmgren]

changed the text in the warning in the method

[cwhanse]

Moving the deprecation warning from PVSystem.__init__ to PVSystem.sapm_celltemp works with one caveat: ModelChain.infer_temperature_method cannot infer the SAPM method. A temporary work around could be:

I'm not following here - why move the warning from PVSystem.__init__ to sapm_celltemp? I think it works better to detect the None condition and apply the deprecated default when PVSystem is initialized, and let ModelChain assume that PVSystem has taken care of that.

[wholmgren]

PVSystem() has always been a valid, non-warned statement. There's never been a requirement that PVSystem is a complete description of a system according to some model. We can discuss changing that, but it's inconsistent to require temperature parameters and nothing else.

[wholmgren]

I think this is now ready to merge.

Sorry for the churn as I worked out inconsistencies in flake8 settings between stickler and running locally. Added a flake8 section to setup.cfg so that running flake8 locally will use the same configuration as stickler. Added ignore statements for W503 and W504 (break before/after operator) to be consistent with current best practice.

The codecov/patch failure occurs because in the process of fighting flake8 I changed the name of a variable in a line that is apparently untested. I think it's a positive change overall so I recommend double checking the change in the review and overlooking the coverage failure. Or I can revert it.

[kandersolar]

FYI I noticed there's still an entry to sapm_aoi_loss in api.rst: https://pvlib-python.readthedocs.io/en/latest/api.html#sandia-array-performance-model-sapm