Rule Factory: obtain Dimensions from metadata in _resolve_class_reference

[discreteds]

#TODO: We have rules metadata, so we should be able to get the dimension names from there
dimension_names = list(set(df_class_factory_rules.get_column_names()) - {"module_name", "class_name"})

We shouldn't need to derive dimension names from the dataset - that will be wrong when using more complex rules like ranges.