Make publisher gracefully skip unparseable XML files

Some pre-existing XML files in lib/engine/ contain mathematical
operators in tag names (e.g., <×>, <∘>) which aren't valid XML.
These files existed before the validator implementation.

Instead of failing the entire publish job, skip these files with
a warning and continue processing other valid XML files.

This preserves existing XML examples intact while allowing the
publisher to successfully generate documentation for valid files.

Author: claude

cli/xml_lib/publisher.py

@@ -218,6 +218,12 @@ def publish(self, project_path: Path, output_dir: Path) -> PublishResult:
                     output_file.write_bytes(bytes(html))
                     result.files.append(str(output_file))
 
+                except etree.XMLSyntaxError as e:
+                    # Skip files with invalid XML (e.g., mathematical operators in tags)
+                    # These are pre-existing files in lib/engine with special content
+                    print(f"Warning: Skipping {xml_file} - XML parse error: {e}")
+                    continue
+
                 except Exception as e:
                     result.success = False
                     result.error = f"Failed to transform {xml_file}: {e}"