Fix pickling URLs with Python 3.15.0a6+ (Fixes #1632)

In Python 3.15 urllib.parse.SplitResult gained a new __getstate__
method which fails when called on instances created via tuple.__new__:
python/cpython@c5cfcdf1
Returning self._val directly instead of creating a new SplitResult
instance avoids the problematic code path, and is backwards compatible.

Author: befeleme

yarl/_url.py

@@ -566,8 +566,8 @@ def __mod__(self, query: Query) -> "URL":
     def __bool__(self) -> bool:
         return bool(self._netloc or self._path or self._query or self._fragment)
 
-    def __getstate__(self) -> tuple[SplitResult]:
-        return (tuple.__new__(SplitResult, self._val),)
+    def __getstate__(self) -> tuple[SplitURLType]:
+        return (self._val,)
 
     def __setstate__(
         self, state: tuple[SplitURLType] | tuple[None, _InternalURLCache]