v0.0.26

Fix: Entity-Decoded Syntax Highlighting

Fixes embedded language tokenization in custom code tags (e.g., <pl-code language="c">).

Bug: The tree-sitter parser produces separate text and html_entity children for entity-encoded content — not a single text node. The v0.0.25 content extraction looked for html_text children which don't exist, so no embedded tokens were ever generated.

Fix: Extract content by slicing the source text between the start tag end and end tag start positions, which captures all content including entities and whitespace regardless of how the parser splits children.