[JammingBen]

Description

Raw HTML in markdown files was lost on save. Tiptap parsed it through the editor schema, so unknown tags kept only their text and comments disappeared.

• Keep raw HTML: the markdown editor now keeps raw HTML as source text and saves it unchanged.
  • Blocks (e.g. <div align="center">, <details>, comments) become an htmlBlock node. It looks like a code block with an "HTML" label.
  • Inline tags (e.g. <kbd>, <sup>, inline comments) become an htmlInline mark, so surrounding formatting stays intact.
  • The editor's own output still goes through the schema as before: resized <img> tags and <br> hard breaks.
• Decode named entities: named HTML entities (e.g. ©,  ) are now decoded. Before, they were shown literally and escaped again on save.
• Viewers: read-only viewers (TextEditorViewer) use the new renderHtml option. They still render HTML through the schema instead of showing its source.

Known limitations

#4026

[kulmann]

Explorative testing was fine and code looks good. One UX nit: let keyboard users exit the html block via Shift+Enter and triple-Enter like in our code block.