Remove redundant agent graceful restart

[DavidePrincipi]

The agent no longer needs a restart during module updates since its configuration is no longer dependent on the module itself. A restart is only required during core updates.

I noticed SIGUSR1 runs multiple times during core-update if core and Loki (for example) are updated together.

[root@rl1 ~]# journalctl --since=07:03:19 --grep Signal
Sep 13 07:03:19 rl1.dp.nethserver.net agent@node[31796]: Signal "user defined signal 1" caught: shutdown started.
Sep 13 07:03:19 rl1.dp.nethserver.net agent@ldapproxy1[32757]: Signal "user defined signal 1" caught: shutdown started.
Sep 13 07:03:19 rl1.dp.nethserver.net agent@loki1[32786]: Signal "user defined signal 1" caught: shutdown started.
Sep 13 07:03:19 rl1.dp.nethserver.net agent@traefik1[32152]: Signal "user defined signal 1" caught: shutdown started.
Sep 13 07:03:27 rl1.dp.nethserver.net agent@loki1[33746]: Signal "user defined signal 1" caught: shutdown started.
Sep 13 07:03:42 rl1.dp.nethserver.net agent@cluster[31791]: Signal "user defined signal 1" caught: shutdown started.

The removed lines were originally added in this commit a90baa3.

When SIGUSR1 is caught, the agent stops running new tasks until the currently running ones finish. This can lead to deadlocks if a running task enqueues a new subtask and waits for it. I think we need to modify the current behavior to prevent these deadlocks (in another PR).

Refs NethServer/dev#7016