Ignore invalid attributes when scraping

[fbacall]

When scraping resources, the various model validations can sometimes be overly strict.

For example when parsing a Bioschemas LearningResource, if a resource is marked-up with:

"educationalLevel": "Introductory"

which TeSS uses for the difficulty_level field, it will fail validation because TeSS uses a controlled vocab for this field and "Introductory" is not a valid option. This validation failure prevents the material being registered entirely.

Instead, TeSS could just ignore that one field and register the material anyway.

• Look into a way of stripping invalid attributes from a model and re-attempting the creation
• Ensure these "soft failures" are logged somewhere as warnings (in the source's ingestion log)