First of all thank you for posting these mishaps and the work around.
We didn't encounter these problems when upgrading from sp2 cu2 to SP3 on our 2-node sql-cluster (multiple instances).
We do pay attention for the cluster resource to reside on the same node as the sql instance to be upgraded.
We even upgraded multiple instances in a single run. We moved all instances to the same node before starting the upgrade.
Off course it will not hurt to have a spare/backup set of resource db files !
We take a copy of them after every upgrade (because at that moment is planned downtime).
One thing we did observer is the needed reboot after the first instance + the client software have been upgraded.