fix(longevity_test): Set ignore raft topology connection close error globally

Error: 'raft_topology - topology change coordinator fiber got error  (connection is closed))'
Could appeared in different moment when node/nodes are being restarted.
As described in issue: scylladb/scylladb#20950 in comment:
scylladb/scylladb#20950 (comment)

the error message could appeared while we have race between raft and gossip.
we can ignore this issue while issue will not be fixed on scylla side,
gossip mode will be removed from scylla.

Author: aleksbykov

sdcm/sct_events/setup.py

@@ -33,6 +33,7 @@
     EVENTS_GRAFANA_ANNOTATOR_ID, EVENTS_GRAFANA_AGGREGATOR_ID, EVENTS_GRAFANA_POSTMAN_ID, \
     EventsProcessesRegistry, create_default_events_process_registry, get_events_process, EVENTS_HANDLER_ID, EVENTS_COUNTER_ID
 from sdcm.utils.issues import SkipPerIssues
+from sdcm.sct_events.group_common_events import ignore_topology_change_coordinator_errors
 
 
 EVENTS_DEVICE_START_DELAY = 1  # seconds
@@ -148,5 +149,12 @@ def enable_default_filters(sct_config: SCTConfiguration):  # pylint: disable=unu
                                 event_class=DatabaseLogEvent.ABORTING_ON_SHARD,
                                 regex=r'.*Parent connection [\d]+ is aborting on shard').publish()
 
+    # As written in https://github.com/scylladb/scylladb/issues/20950#issuecomment-2411387784
+    # raft error messages with connection close could be ignored during topology operations,
+    # upgrades and any place where the race between raft global barrier and gossipier could
+    # take place. So ignore such messages globally for any sct test.
+    # TODO: this should be removed after gossiper will be removed.
+    ignore_topology_change_coordinator_errors().__enter__()
+
 
 __all__ = ("start_events_device", "stop_events_device", "enable_default_filters")