Make a 'sanity check' test to check that the default model returns reasonable results

[nickmalleson]

No description provided.

[github-actions]

Branch nickmalleson-issue-232 created!

[nickmalleson]

Notes about what we might do first (but maybe redundant if we fix #231 ):

• Move 'gam_cases.csv' to 'devon_data' (or anywhere else really; it shouldn't hang in the root directory)
• Add a parameter in default.yml (observations-file) to specify where that file can be found

Started to make a test, but then gave up. Here's some code I'll put back in net year:

def test_model_performance(setup_results):
    """This is a high-level test that checks whether the model is performing
    reasonably well against a benchmark of real observations. Useful if a change
    means that the default model no longer works properly."""
    #TODO Implement this