Fix nvidia_gpu_enabled breaking helm deployment

Two problems where found:

• The daemonset to prelabel the gpu nodes with the mig/config: disabled is needed (we dropped it)
• Drop restartPolicy on seLinux daemonset. Without this the selinuc daemonset installation is halting the helm chart installation due to: Error: DaemonSet.apps "nvidia-selinux-disable" is invalid: spec.template.spec.restartPolicy: Unsupported value: "OnFailure": supported values: "Always"