fix: link validation doesn't ignore fenced code blocks when \r is present

Signed-off-by: squidfunk <martin.donath@squidfunk.com>

python/tests/unit/collectors/test_references.py

@@ -1470,6 +1470,14 @@ class TestFencedCodeBlocks:
                 b"~~~~ py\n[text](href)\n~~~~",
                 id="fenced-code-4-tildes-lang",
             ),
+            pytest.param(
+                b"\n```\n[Start][]\n```\n",
+                id="fenced-code-with-info",
+            ),
+            pytest.param(
+                b"```\r\n[Start]\r\n```\r\n",
+                id="fenced-code-crlf-with-shortcut-link-ref",
+            ),
         ],
     )
     def test_fenced_code(self, md: bytes) -> None:

python/zensical/collectors/references/cursor.py

@@ -971,7 +971,7 @@ def _scan_fenced_code(cursor: Cursor, char: int) -> int | None:
         # Skip to next line and terminate if we found a closing fence
         if end - start == opening:
             rest = _skip_whitespace(cursor, end)
-            if rest >= cursor.end or cursor.data[rest] == _NL:
+            if rest >= cursor.end or cursor.data[rest] in (_CR, _NL):
                 return _skip_line(cursor, end)
 
         # Skip to next line