✨ feat(sanitize): rename tags during sanitize

[gaborbernat]

Sanitizing could drop or keep an element but not rewrite it, so migrating a sanitize-html config that renames deprecated presentational tags (<b>→<strong>, <center>→<div>) had no equivalent. This adds Policy.transform_tags, the port of sanitize-html's transformTags, so a rename happens inside the single sanitize pass rather than as a second edit over the tree.

Map a source tag to a string to rename it, or to a Transform to rename it and add attributes (sanitize-html's simpleTransform). The rename runs at the top of the element walk, before the allowlist reads the tag, and the walk then continues on the renamed element as if the author had written the target. 🔒 That ordering is the safety argument: the allowlist still decides the element's disposition, the unconditional baseline still escapes a script or iframe target, and any injected attribute joins the element's own to be scrubbed by the same gates. {"b": "script"} cannot smuggle a live <script>, and an injected href cannot carry a javascript: URL. Only HTML elements are transformed, matched by tag name.

The rename mutates the node in place through the existing tree primitives (rewrite the name, relookup the atom, preserve the source-closed flag), so a transform adds no extra pass over the document. The migration guide benchmarks the transform op end-to-end against sanitize-html's Node runner over stdin, the cost a Python service pays to reach it as a subprocess.

closes #531

[codspeed-hq]

Merging this PR will not alter performance

✅ 66 untouched benchmarks
🆕 1 new benchmark
⏩ 18 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
🆕	test_feature[sanitize-transform-spec]	N/A	12.7 ms	N/A

---

_{Comparing gaborbernat:feat/531-transform-tags (65e05a3) with main (e9032d7)}

Footnotes

1. 18 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩