tasks: Run all cockpit-tasks@* instances in a systemd slice

With this we can react to all tasks containers shutting down. It also
makes it easier to stop all instances without having to resort to
globbing or knowing how many instances a host has.

Note: Avoid a `-` in the name, as that will create a
sub-directory/slice.

tasks/install-service

@@ -37,6 +37,7 @@ Requires=podman.socket
 After=podman.socket
 
 [Service]
+Slice=cockpittasks.slice
 Restart=always
 RestartSec=60
 # give image pull enough time