Fix: crmd: update crm_peer_cache
This should make crm_node -l working correctly after restarting a node