#648 update cmor version to v3-13-0

[mo-jareddrayton]

This closes #648

CMOR Update / Environment Changes

• This was mostly straightforward other than having to unpin nccmp to get the environment to resolve. This resulted in the nccmp version changing from 1.9.1 -> 1.10.0 which I wouldn't expect to cause any problems being a minor version change. All other version changes in unpinned packages look reasonable.

Code Changes

Most of the awkwardness with this change come from the change in variant label and how the usage of cmor_dataset_config provides the function that does the validation.

CDDS/mip_convert/mip_convert/configuration/python_config.py

Line 276 in 93029c6

_configs = [cmor_setup_config(), cmor_dataset_config(), request_config()]

Without redesigning the current implementation we need to provide this function with the mip_era information. As a class variable I can't see any obvious way to do this so I don't see any drawbacks with making _configs an instance variables which is assigned on initialisation and can be provided the correct context information.

CDDS/mip_convert/mip_convert/configuration/python_config.py

Lines 306 to 307 in f53efd3

	def _add_configs(self):
	self._configs = [cmor_setup_config(), cmor_dataset_config(self.mip_era), request_config()]

There is a simpler solution that I tried first which removes the need for adding the mip_era argument to the Class and having to pass the plugin_id through many functions you can just use self.config["cmor_dataset]["mip_era"] within the UserConfig class. the problem is this causes some (somewhat minor) tests to fail. In hindsight I'm almost tempted to revert back to my initial solution and try and save what tests I can.

Example breaking test. You need the "cmor_dataset to get the "mip_era"!

CDDS/mip_convert/mip_convert/tests/test_configuration/test_python_config.py

Lines 115 to 121 in f53efd3

	def test_missing_cmor_setup_section(self, mopen):
	user_config = '[cmor_dataset]\n[request]\n'
	mopen.return_value = StringIO(dedent(user_config))
	msg = 'User configuration file does not contain the required section "cmor_setup"'
	self.assertRaisesRegex(ValidateConfigError, msg, UserConfig, self.read_path, self.history)
	mopen.assert_called_once_with(self.read_path)

[mo-jareddrayton]

One other thing I noticed during development, if I remove model_type I get these cmor errors.

CDDS/mip_convert/mip_convert/tests/test_functional/utils/configurations.py

Line 223 in c420dfd

'model_type': 'AOGCM AER',

I wasn't sure if this is a bug in one of cmor / CV.json / mip_convert.

!!!!!!!!!!!!!!!!!!!!!!!!!
!
! Error: The following source type(s) "AOGCM " are required and
! some source type(s) could not be found in your input file. 
! Your file contains a source type of "".
! Check your Controlled Vocabulary file "$CDDS_ETC/mip_tables/CMIP7/DR-1.2.2.2-v1.0.1/CMIP7_CV.json".
! 
!
!!!!!!!!!!!!!!!!!!!!!!!!!


C Traceback:
! In function: _CV_checkSourceType
! called from: _CV_checkExperiment
! 

!!!!!!!!!!!!!!!!!!!!!!!!!
!
! Error: You source_type attribute contains invalid source types
! Your source type is set to "".  The required source types
! are "AOGCM " and possible additional source types are "AER CHEM BGC " 
! Check your Controlled Vocabulary file "$CDDS_ETC/mip_tables/CMIP7/DR-1.2.2.2-v1.0.1/CMIP7_CV.json".

Is this expected @matthew-mizielinski ?

[matthew-mizielinski]

Is this expected @matthew-mizielinski ?

I think this is a bug in CMOR, but is resolvable by removing the additional_allowed_model_components and required_model_components entries from the CV. My fault for not noticing this when putting this file in place.

[matthew-mizielinski]

LGTM and all tests pass