We've seen handlers being inconsitently being ignored for some
deployments when we force a handler flush mid-playbook. We'll need to
ensure the config is properly loaded if we want to follow up with
subsequent chronyc application so it's best to always invoke the restart
when the configuration actually changes.
Change-Id: I5292f1b1bd65cef340e1e9742463620effc30d26
068668b4bb