TestEvaluate test must be run from python/lib/evaluations

[aaraney]

python/lib/evaluations/dmod/test/test_evaluate.py::TestEvaluate tests must be run from python/lib/evaluations or the tests fail.

Simple fix is changing to:

    def setUp(self) -> None:
        self.oldwd = os.getcwd()
        # from root, this would be: `python/lib/evaluations`
        evaluations = Path(__file__).parent.parent.parent
        os.chdir(evaluations)
        self.__cfs_to_cms_specification = self.get_cfs_to_cms_specification()
        self.__cfs_to_cfs_specification = self.get_cfs_to_cfs_specification()

    def tearDown(self) -> None:
        os.chdir(self.oldwd)

Not sure if this is exactly what we want or if this is more systemic. @christophertubbs, can you weigh in on this?

Edit: Changing the current working directory is necessary b.c. evaluate.Evaluator specification's can include backends (external resources) via relative or absolute file paths (see linked example). Tests in this test suite use specifications that define resources with relative file paths that improperly resolve when the tests are run from unexpected working directories.

[aaraney]

You can reproduce this issue by running the test from the root of the repo with python -m unittest python/lib/evaluations/dmod/test/test_evaluate.py.