Class stripping removes language-* from code blocks, losing the code language #34

Problem

LoopHtmlConverter.Convert strips every class attribute from the rendered HTML:

// Remove CSS classes that Markdig adds
html = System.Text.RegularExpressions.Regex.Replace(html, @"\s*class=""[^""]*""", "");

Markdig renders a fenced block with its language as <code class="language-cs">. The regex deletes that, so the language is gone from the output HTML.

Repro

Input Markdown:

```cs
var x = 1;
```

Output HTML:

<pre><code>var x = 1;
</code></pre>

Round-tripping back through HtmlToMarkdownConverter therefore yields an unlabelled fence — the cs hint is lost permanently.

Impact

• Code language is dropped on every Markdown → Rich Text conversion, so Loop cannot syntax-highlight and the information cannot be recovered.
• The regex is broader than its stated intent ("classes that Markdig adds") — it also strips classes from any raw HTML the user deliberately wrote in their Markdown.

Suggested fix

Preserve language-* classes and strip only what Loop actually chokes on. Either:

• narrow the regex so class="language-…" is retained, or
• drop the regex entirely and configure the renderer not to emit the unwanted classes in the first place (cleaner, and consistent with the renderer-based approach proposed in TransformTaskLists is dead code; checkbox output relies on brittle string replacement #21).

Found while verifying #16.