Add warning-feature to DataValidator

[danielhuppmann]

As a next step for scenario validation and vetting, we need to extend the DataValidator class to enable showing a warning instead of raising an error.

My proposed implementation is the following:

• Add a warning-level attribute to a validation-criteria item, which defaults to None, but can be set to e.g. "red", "yellow"
• This will probably require an intermediate class DataValidationCriteria(IamcDataFilter) with the new attribute, and DataValidationCriteriaValue and DataValidationCriteriaBounds inherit from that.

• If the validation fails and the warning-level is not None, add the value to a warning-list in https://github.com/IAMconsortium/nomenclature/blob/3a76cba2f5baf6c7f54be9d45d40c1a698738f6c/nomenclature/processor/data_validator.py#L96

  instead of the error-list.

• If the warning-list is not empty, display the warning-list as a DataFrame (after the error-list, but before raising the error).

Does that make sense, @phackstock?

[dc-almeida]

Clarification:

display the warning-list as a DataFrame

Any example of how this would look? And from what I understand, the fallback is always on raising the error if no warning level is specified.

[danielhuppmann]

Any example of how this would look?

I'd say as a first step, take the same as

https://github.com/IAMconsortium/nomenclature/blob/3a76cba2f5baf6c7f54be9d45d40c1a698738f6c/tests/test_validate_data.py#L108

but add a "warning-level" column.

And from what I understand, the fallback is always on raising the error if no warning level is specified.

Correct.