Clusters with large numbers of CSVs can cause crashloop, block upgrades

A cluster was upgrading from 4.12.9 toward 4.12.17 but the upgrade was stuck during the upgrade of the node-tuning-operator.

The reason is that the node-tuning-operator is in a crashloop, and this appears to be because it is trying to get a list of CSVs, and this takes too long, causing a timeout:

$ oc get pods -n openshift-cluster-node-tuning-operator
NAME                                            READY   STATUS             RESTARTS         AGE
cluster-node-tuning-operator-7879747fc9-x8fqf   0/1     CrashLoopBackOff   33 (3m40s ago)   3h7m

$ oc logs -f cluster-node-tuning-operator-7879747fc9-x8fqf -n openshift-cluster-node-tuning-operator
...
unable to remove Performance addons OLM operator: the server was unable to return a response in the time allotted, but may still be processing the request (get clusterserviceversions.operators.coreos.com)
...

The cluster has 114 CSVs

4.12.9 (upgrading from)
4.12.17 (upgrading to)

Reproducible during upgrade

Upgrade stays stuck

Upgrade should not be stuck