cleanup action leaves failcounts behind

[Xav Paice]

On some, but not all, occasions when a resource failcount is >1, the cleanup action does not clear all the failcounts and we end up needing to run something along the following lines to clean the resources:

for i in $(jsft |grep hacluster |awk '/ubuntu/ {print $1}') ; do juju run --application $i 'sudo crm_resource --list -Q -l | xargs -L1 sudo crm_failcount -D -r' ; done

Note that the crm_failcount -D -r needs to be run on each unit of an application in order to be effective, though the crm_resource -C is cluster wide.

I suggest we add `crm_failcount -D` to the action, for each resource on the unit. In order to ensure the whole cluster is clear, we might need to run the action on all units, but at least we can do so via actions rather than 'juju run' or ssh (which requires different permissions for some models of managed services).