fix(nemesis): increase timeout waiting node down

After logs investigation, wait of node to be down
by nemesises: disrupt_refuse_connection_with_* could take
more than 10 minutes for test configuration with
tablets, large data sets and multidc clusters, especially
when verification node and target nodes are in different DCes

Fixes: #10434

Author: aleksbykov

sdcm/nemesis.py

@@ -5522,18 +5522,19 @@ def drop_keyspace(node):
         simulate_node_unavailability = node_operations.block_scylla_ports if use_iptables else node_operations.pause_scylla_with_sigstop
         with self.run_nemesis(node_list=self.cluster.nodes,
                               nemesis_label=f"Running {simulate_node_unavailability.__name__}") as working_node, ExitStack() as stack:
+            stack.callback(drop_keyspace, node=working_node)
             target_host_id = self.target_node.host_id
             stack.callback(self._remove_node_add_node, verification_node=working_node, node_to_remove=self.target_node,
                            remove_node_host_id=target_host_id)
 
             self.tester.create_keyspace(keyspace_name, replication_factor=3)
             self.tester.create_table(name=table_name, keyspace_name=keyspace_name, key_type="bigint",
                                      columns={"name": "text"})
-            stack.callback(drop_keyspace, node=working_node)
 
             with simulate_node_unavailability(self.target_node):
-                # target node stopped by Contextmanger. Wait while its status will be updated
-                wait_for(node_operations.is_node_seen_as_down, timeout=600, throw_exc=True,
+                # target node stopped by Contextmanger. Wait while its status will be updated.
+                # with tablets and multidc it could take more time.
+                wait_for(node_operations.is_node_seen_as_down, step=10, timeout=3600, throw_exc=True,
                          down_node=self.target_node, verification_node=working_node, text=f"Wait other nodes see {self.target_node.name} as DOWN...")
                 self.log.debug("Remove node %s : hostid: %s with blocked scylla from cluster",
                                self.target_node.name, target_host_id)
@@ -5560,12 +5561,11 @@ def drop_keyspace(node):
 
             with self.cluster.cql_connection_patient(working_node) as session:
                 LOGGER.debug("Check keyspace %s.%s is empty", keyspace_name, table_name)
-                result = list(session.execute(f"SELECT * from {keyspace_name}.{table_name}"))
+                result = list(session.execute(
+                    f"SELECT * from {keyspace_name}.{table_name}", consistency_level=ConsistencyLevel.QUORUM))
                 LOGGER.debug("Query result %s", result)
                 assert not result, f"New rows were added from banned node, {result}"
 
-            drop_keyspace(working_node)
-
 
 def disrupt_method_wrapper(method, is_exclusive=False):  # pylint: disable=too-many-statements  # noqa: PLR0915
     """