Child extractors are now directly constructed with Extractor.from_url() if the extractor class is known beforehand, instead of using extractor.find() and searching through all possible extractor classes.