Stands in for youtube.com so the DOM-harvest strategy can be exercised: the transcript loads lazily in pages of 20, and every custom element has been renamed. Nothing here matches the tag names the extractor prefers, so only the structural fallback can find the transcript.