Serverless operator upgrade fails with "Service account is owned by another ClusterServiceVersion". This error can be seen on the ClusterServiceVersion resource.

4.12.15, 4.10.59

This error is intermittent. It's probably a race condition. A similar error was reported in the past: https://bugzilla.redhat.com/show_bug.cgi?id=1934080

1. Create namespace and operatorgroup for the operator:

oc create namespace openshift-serverless
cat <<EOF | oc apply -f -
apiVersion: operators.coreos.com/v1
kind: OperatorGroup
metadata:
  name: serverless
  namespace: openshift-serverless
EOF

2. Install the previous version of Serverless operator (version 1.27):

cat <<EOF | oc apply -f -
apiVersion: operators.coreos.com/v1alpha1
kind: Subscription
metadata:
  name: serverless-operator
  namespace: openshift-serverless
spec:
  channel: stable
  name: serverless-operator
  source: redhat-operators
  sourceNamespace: openshift-marketplace
  installPlanApproval: Manual
  startingCSV: serverless-operator.v1.27.0
EOF

3. Approve the CSV via UI or by approving the given InstallPlan in the openshift-serverless namespace.
4. Then, wait for the Serverless 1.28 to be offered, perform the upgrade by approving the CSV again.

The ClusterServiceVersion shows this status:
"requirementStatus": [
                   ...
                    {
                        "group": "",
                        "kind": "ServiceAccount",
                        "message": "Service account is owned by another ClusterServiceVersion",
                        "name": "knative-openshift",
                        "status": "PresentNotSatisfied",
                        "version": "v1"
                    },
                    {
                        "group": "",
                        "kind": "ServiceAccount",
                        "message": "Service account is owned by another ClusterServiceVersion",
                        "name": "knative-operator",
                        "status": "PresentNotSatisfied",
                        "version": "v1"
                    },
                    {
                        "group": "",
                        "kind": "ServiceAccount",
                        "message": "Service account is owned by another ClusterServiceVersion",
                        "name": "knative-openshift-ingress",
                        "status": "PresentNotSatisfied",
                        "version": "v1"
                    }
                ]

The ClusterServiceVersion is ready, without errors.

We have seen this error on both OCP 4.10 and 4.12. It happens in the scenario described above when performing a first upgrade. This was seen e.g. here on OCP 4.10:
https://prow.ci.openshift.org/view/gs/origin-ci-test/logs/periodic-ci-openshift-knative-serverless-operator-main-4.10-upgrade-tests-aws-ocp-410-continuous/1655392361728970752
(I'm attaching must-gather logs from this run)

It also happens (more often) during a second upgrade (after performing downgrade). Examples:
OCP 4.10:
https://prow.ci.openshift.org/view/gs/origin-ci-test/logs/periodic-ci-openshift-knative-serverless-operator-main-4.10-upgrade-tests-aws-ocp-410-continuous/1657385277863235584
OCP 4.12
https://prow.ci.openshift.org/view/gs/origin-ci-test/logs/periodic-ci-openshift-knative-serverless-operator-main-4.12-upgrade-tests-aws-ocp-412-continuous/1657928885674708992

When performing downgrade the steps from the reproducer are performed twice, but in the meantime the operator is completely uninstalled using these steps:
* Delete Subscription
* Delete ClusterServiceVersion
* Delete the operator namespace openshift-serverless