Description
We should have a standard way to indicate how a schema 'model' is described. For instance, Parquet models use a DSNP-centric JSON description format (enables bloom filters, but very limited in other ways), but other JSON formats exist that support different Parquet features. ModelType::Parquet doesn't indicate how to interpret the model.
Also: the chain currently validates Parquet models at the time of schema publication, but only validates against the "DSNP JSON".