fix: restart Containerd when setting mirror config

[dkoshkin]

Updates under this directory do not require restarting the containerd daemon.

Even-though a restart is not required on changes, it is still required initially after setting config_path in the Containerd config.

---

Tested locally with a CAPD cluster:

root@docker-quick-start-helm-addon-cilium-ng7q8-7xcrh:/# journalctl -u containerd | grep starting
Feb 16 01:42:07 docker-quick-start-helm-addon-cilium-ng7q8-7xcrh containerd[129]: time="2024-02-16T01:42:07.338020972Z" level=info msg="starting containerd" revision=1677a17964311325ed1c31e2c0a3589ce6d5c30d version=v1.7.1
Feb 16 01:42:17 docker-quick-start-helm-addon-cilium-ng7q8-7xcrh containerd[495]: time="2024-02-16T01:42:17.495879170Z" level=info msg="starting containerd" revision=1677a17964311325ed1c31e2c0a3589ce6d5c30d version=v1.7.1

Also tested in an AWS instance with a globalMirrorRegistry configuration and after restarting Containerd, the Kubelet was able to start Pods.