rsyslog step in juju-restore is outdated and breaks the restore

Andrew Wilkins said:

We no longer write rsyslog configuration at bootstrap time, but instead it's managed by a worker in jujud.
The upshot is we should be able to just remove anything to do with rsyslog from the restore plugin. It should fix itself up when the agent starts up. Note this is only true for >=1.18, so we may want a version check in there.

In current trunk version we get upon attempting a restore:

+ cd /var/lib/juju/agents
+ for agent in '*'
+ initctl stop jujud-machine-1
jujud-machine-1 stop/waiting
+ sed -i.old -r '/^(stateaddresses|apiaddresses):/{
                n
                s/- .*(:[0-9]+)/- ec2-54-205-27-132.compute-1.amazonaws.com/
        }' machine-1/agent.conf
+ [[ machine-1 = unit-* ]]
+ initctl start jujud-machine-1
jujud-machine-1 start/running, process 5101
jujud-unit-ubuntu-0 stop/waiting
jujud-unit-ubuntu-0 start/running, process 5106
+ for agent in '*'
+ initctl stop jujud-unit-ubuntu-0
+ sed -i.old -r '/^(stateaddresses|apiaddresses):/{
                n
                s/- .*(:[0-9]+)/- ec2-54-205-27-132.compute-1.amazonaws.com/
        }' unit-ubuntu-0/agent.conf
+ [[ unit-ubuntu-0 = unit-* ]]
+ sed -i -r 's/change-version: [0-9]+$/change-version: 0/' 'unit-ubuntu-0/state/relations/*/*'
sed: can't read unit-ubuntu-0/state/relations/*/*: No such file or directory
+ true
+ initctl start jujud-unit-ubuntu-0
+ sed -i -r 's/^(:syslogtag, startswith, "juju-" @)(.*)(:[0-9]+.*)$/ec2-54-205-27-132.compute-1.amazonaws.com/' '/etc/rsyslog.d/*-juju*.conf'
sed: can't read /etc/rsyslog.d/*-juju*.conf: No such file or directory

This bug is being fixed on under the context of #1291022