Use builtins.TimeoutError consistently in cluster probe drain

cluster.py:546 was the lone straggler still suppressing
asyncio.TimeoutError while every sibling site (protocol.py,
sqlalchemy-dqlite/base.py) catches the builtins.TimeoutError name
introduced in 3.11. The two are aliased today, so the behaviour is
identical, but the alias has been deprecated in favour of the
builtin and the codebase's source-of-truth comment in protocol.py
already documents the convention.

One-character forward-compat hygiene fix.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: antoineleclair

src/dqliteclient/cluster.py

@@ -543,7 +543,7 @@ async def _query_leader(
             # the drain to completion within its 100 ms budget even
             # during shutdown; the outer cancel still propagates past
             # this ``finally`` as expected.
-            with contextlib.suppress(OSError, asyncio.TimeoutError):
+            with contextlib.suppress(OSError, TimeoutError):
                 await asyncio.shield(
                     asyncio.wait_for(
                         writer.wait_closed(),