controlernode left shmem locks behind forcing save_brm to fail b/c of the locks

There is a shell script that kills controllernode after 20 seconds grace period. The grace period exists b/c controllernode tries to gracefully stop its workernode connections. Sometimes it is impossible to stop workernode connection gracefull and killed controllernode leaves shmem locks behind that causes save_brm to stuck and thus nodes doesn't have actual BRM image saved at shutdown.
save_brm.py must check whether there are active shmem locks and clean them if possible.