When the parser encounters a stray `</body>` or `</html>` end tag but the document still has more content, what does tree construction do, and why does subsequent content still attach correctly?