Validation of User Configuration

[pgierz]

This adds a collection of validators that can be run against the user supplied dictionaries to check for errors.

Closes #27

[pgierz]

This looks promising for capturing user errors, just need some nice formatting:

        if "rules" in data:
            if not RULES_VALIDATOR.validate({"rules": data["rules"]}):
>               raise ValueError(RULES_VALIDATOR.errors)
E               ValueError: {'rules': [{0: [{'description': ['unknown field'], 'enabled': ['unknown field'], 'pipelines': ['unknown field']}], 1: [{'cmor_units': ['unknown field'], 'description': ['unknown field'], 'enabled': ['unknown field'], 'pipelines': ['unknown field']}], 2: [{'cmor_variable': ['null value not allowed'], 'enabled': ['unknown field']}], 3: [{'cmor_variable': ['null value not allowed'], 'pipelines': ['unknown field']}]}]}

[pgierz]

There is one part I am not sure about that @siligam should check.

[pgierz]

@siligam: This still needs to be added correctly for units.py. I wasn't sure how you had designed it.

[siligam]

@pgierz "model_units" is referred as "source_units" in units.py. It is a optional parameter. If it is not provided, it reads the model units from xarray dataset. At times, user prefers to use some other unit instead of units from xarray dataset. In this case, "source_units" takes precedence if provided.
The setting "model_units" : {"required": False} is apt.