Restart gitlab-workhorse when trusted certificates change

If workhorse is not restarted after certificates change in trusted certificates store, it will fail to pick up the new certificates.

How to reproduce

1. Deploy redis externally using a custom signed certificate.
2. Configure GitLab to use this redis instance. DO NOT add the custom CA to the trusted certificates yet.
3. Check the logs of `gitlab-workhorse. You will see something like

   {"level":"info","msg":"keywatcher: starting process loop","time":"2021-12-03T09:35:33Z"}
   {"address":"<REDIS_URL>","level":"info","msg":"redis: dialing","scheme":"rediss","time":"2021-12-03T09:01:53Z"}
   {"error":"keywatcher: x509: certificate signed by unknown authority","level":"error","msg":"","time":"2021-12-03T09:01:53Z"}

4. Add the custom CA to trusted certificates and run reconfigure
5. Workhorse hasn't picked up the changes yet and the errors continue to pop up.
6. Restart workhorse, and see the error message go away

   {"level":"info","msg":"keywatcher: starting process loop","time":"2021-12-03T09:37:57Z"}
   {"address":"<REDIS_URL>","level":"info","msg":"redis: dialing","scheme":"rediss","time":"2021-12-03T09:37:57Z"}