Markdown: add fake base class to Pattern to fix overrides

Currently `InlineProcessor` and every subclass of it violates the `override` rule as per Liskov Substitution Principle.

    Signature of "handleMatch" incompatible with supertype "Pattern"  [override]
         Superclass:
             def handleMatch(self, m: Match[str]) -> Element | str
         Subclass:
             def handleMatch(self, m: Match[str], data: str) -> tuple[Element | str | None, int | None, int | None]

For all *usage* purposes these aren't actual subclasses to one another, they're handled totally separately inside `__applyPattern`.

However yes, this change totally disagrees with actual `isinstance` checks at run time (though even that is still for the best).

Author: oprypin

stubs/Markdown/markdown/inlinepatterns.pyi

@@ -39,21 +39,23 @@ class EmStrongItem(NamedTuple):
     builder: str
     tags: str
 
-class Pattern:
+class _BasePattern:
     ANCESTOR_EXCLUDES: ClassVar[Collection[str]]
     pattern: str
     compiled_re: re.Pattern[str]
     md: Markdown
     def __init__(self, pattern: str, md: Markdown | None = None) -> None: ...
     def getCompiledRegExp(self) -> re.Pattern[str]: ...
-    def handleMatch(self, m: re.Match[str]) -> str | Element | None: ...
     def type(self) -> str: ...
     def unescape(self, text: str) -> str: ...
 
-class InlineProcessor(Pattern):
+class Pattern(_BasePattern):
+    def handleMatch(self, m: re.Match[str]) -> str | Element | None: ...
+
+class InlineProcessor(_BasePattern):
     safe_mode: bool
     def __init__(self, pattern: str, md: Markdown | None = None) -> None: ...
-    def handleMatch(self, m: re.Match[str], data) -> tuple[Element, int, int] | tuple[None, None, None]: ...  # type: ignore[override]
+    def handleMatch(self, m: re.Match[str], data: str) -> tuple[Element, int, int] | tuple[None, None, None]: ...
 
 class SimpleTextPattern(Pattern): ...
 class SimpleTextInlineProcessor(InlineProcessor): ...