csv: add a return type for things that return (#3604)

Technically, these return whatever the write object of the underlying
file object return. I'm not sure it's worth making everything generic
over this probably rarely used return type, but happy to do it (or drop
this change) if that seems better.

stdlib/2and3/_csv.pyi

@@ -32,10 +32,10 @@ class _writer:
     dialect: Dialect
 
     if sys.version_info >= (3, 5):
-        def writerow(self, row: Iterable[Any]) -> None: ...
+        def writerow(self, row: Iterable[Any]) -> Any: ...
         def writerows(self, rows: Iterable[Iterable[Any]]) -> None: ...
     else:
-        def writerow(self, row: Sequence[Any]) -> None: ...
+        def writerow(self, row: Sequence[Any]) -> Any: ...
         def writerows(self, rows: Iterable[Sequence[Any]]) -> None: ...
 
 

stdlib/2and3/csv.pyi

@@ -96,8 +96,11 @@ class DictWriter(object):
         *args: Any,
         **kwds: Any,
     ) -> None: ...
-    def writeheader(self) -> None: ...
-    def writerow(self, rowdict: _DictRow) -> None: ...
+    if sys.version_info >= (3, 8):
+        def writeheader(self) -> Any: ...
+    else:
+        def writeheader(self) -> None: ...
+    def writerow(self, rowdict: _DictRow) -> Any: ...
     def writerows(self, rowdicts: Iterable[_DictRow]) -> None: ...
 
 class Sniffer(object):