[johnhldiver-beep]

Fixes #283.

document.body loses its children: parseHTML('<body><div class="whatever"></div></body>') makes the <body> element itself the documentElement, and the shortcut getters in esm/html/document.js then misfire: head prepends a new empty <head> inside the real body, and body creates a new empty <body> after it. Result: two <body> elements, document.body is the fake empty one, and merely accessing it mutates the real body.

Fix (root cause, one concern): the head/body getters now return documentElement directly when it already is HEAD/BODY, before any create/prepend/after logic. Both mirrors updated identically. title untouched (not affected by this root cause). No existing tests modified.

Tests: regression tests in test/html/document.js for the body case and the symmetric head case. New tests failed before the fix and pass after; test/html/document.js and the full node test/index.js suite both exit 0 with no failures. Repros verified in both ESM and CJS: body === document.body true, exactly 1 body, children still just DIV (no mutation), bare-<head> doc head identity true, and a normal full document (head/body/title) unaffected. Honest limits: this suite prints no numeric assertion totals, so counts are reported as exit codes + failure-grep counts; eslint/c8 were not run (devDeps deliberately skipped for disk).

Notes: prior PR #108 (2022, "Fix document.body not including content from DOMParser parsed ") was closed unmerged because the maintainer incorporated a related fix for #106 directly — and his stated guidance there ("The body getter is likely the place to look for a change") is exactly the approach taken here. Residual, pre-existing, out of scope: calling document.head directly on a bare-<body> document still prepends an empty <head> inside the body (the old ensure-a-head fallback); the issue's repro is about body, and changing head-on-bare-body semantics would be a separate behavior decision.