Parsing whole episodes can be expensive and we only need a subset to determine if it should be indexed/updated or ignored.