Increase job duration in memory-based-scheduling test

The test sometimes fails because the job finishes before
the assertion on its state being RUNNING. This is most
likely caused by a longer than expected cluster update time.

Increase the job duration by 90 seconds to reduce the
probability of the test failing.

Signed-off-by: Jacopo De Amicis <jdamicis@amazon.it>

Author: jdeamicis

tests/integration-tests/tests/schedulers/test_slurm.py

@@ -2619,7 +2619,7 @@ def _test_slurm_behavior_when_updating_schedulable_memory_with_already_running_j
         submit_command_args={
             "nodes": 1,
             "slots": 1,
-            "command": "srun ./a.out 3000000000 300",
+            "command": "srun ./a.out 3000000000 390",
             "other_options": "-w queue1-st-ondemand1-i1-1",
             "raise_on_error": False,
         }