[RobinDev]

Closes part of #37 (point 3).

HtmlToken::parse() re-parses its own content with the inline rules, so the body of a <script> or <style> element was treated as Markdown:

new Markdown(highlighter: null)->parse('<script>const x = "**a**";</script>')->html;
// <script>const x = "<strong>a</strong>";</script>

CommonMark calls these raw text elements — pre, script, style, textarea, §4.6 type 1 — and never parses their content.

HtmlRule now marks those tokens raw (case-insensitively) and HtmlToken::parse() returns their content untouched. Every other element keeps parsing Markdown inside it, which is Tempest's own behaviour and unrelated to #20:

'<div>**a**</div>' // still <div><strong>a</strong></div>

This is the parsing half of point 3. Escaping literal text (a\*b, a_b_c, quotes and entities in text tokens) is the other half and is not in this PR.

composer test (333 tests), lint and analyse are clean.

[github-actions]

Benchmark Results

Comparison of fix/raw-html-content against main (5e06bea775768b2af19579ec592d17d625bac4ef).

Open to see the benchmark results

No benchmark changes above ±5%.

_{Generated by phpbench against commit 2ce1e2d}