Lower failed connection log level to warning

Since the failed connection is caught and handled, i.e. we
will subsequently reconnect, it probably makes sense to
only emit a warning rather than an error.

This may occur e.g. if a cluster job running a wrapper gets
suspended. On job resumption we encounter some form of failed
connection error (typically StreamLostError/ConnectionResetError
or sometimes AMQPHeartbeatTimeout).

Author: Richard Gildea

src/workflows/transport/pika_transport.py

@@ -1482,7 +1482,7 @@ def _run(self):
                 if connection_counter == 0:
                     logger.error(f"Initial connection failed: {e!r}")
                     break
-                logger.error(f"Connection {connection_counter} failed: {e!r}")
+                logger.warning(f"Connection {connection_counter} failed: {e!r}")
             except BaseException as e:
                 logger.error(f"Connection failed for unknown reason: {e!r}")
                 self._exc_info = sys.exc_info()