[tlq5l]

Fixes #283

parseHTML('<body>...') has no <html>, so the <body> element became the document element. document.head and document.body assume <html> is the root, so they created a new <head> and a new, empty <body> inside the real body.

The parser now adds an implicit <html> when a top level <head> or <body> appears in a full document parse (DOMParser / parseHTML). innerHTML fragment parsing is unchanged. I changed esm, regenerated cjs and worker.js, and added a test in test/html/document.js.

Ran live: parseHTML('<body><div class="whatever"></div></body>') -> document.querySelector('body') === document.body is true, and documentElement.outerHTML is <html><head></head><body><div class="whatever"></div></body></html>