OLM fails to upgrade operators immediately

This bug was initially created as a copy of
Bug #2024048
I am copying this bug because:

This bug was initially created as a copy of
Bug #2002276
I am copying this bug because:
Backporting

Description of problem:
Upgrading descheduler from 4.8 to 4.9 fails, i see that when channel & starting CSV is set i do not see any upgrade starting, jian zhang looked further and below is what he found.

1, Only remove the sub, not the csv, resubscribe the 4.9 one, get the below errors:

• message: 'constraints not satisfiable: @existing/openshift-kube-descheduler-operator//clusterkubedescheduleroperator.4.8.0-202108312109
  and qe-app-registry/openshift-marketplace/4.9/clusterkubedescheduleroperator.4.9.0-202109071344
  originate from package cluster-kube-descheduler-operator, subscription cluster-kube-descheduler-operator
  requires qe-app-registry/openshift-marketplace/4.9/clusterkubedescheduleroperator.4.9.0-202109071344,
  subscription cluster-kube-descheduler-operator exists, clusterserviceversion
  clusterkubedescheduleroperator.4.8.0-202108312109 exists and is not referenced
  by a subscription'
  reason: ConstraintsNotSatisfiable
  status: "True"
  type: ResolutionFailed

2,, Remove the sub and csv, and then recreate it, it works well.
[cloud-user@preserve-olm-env jian]$ cat sub-descheduler.yaml
apiVersion: operators.coreos.com/v1alpha1
kind: Subscription
metadata:
name: cluster-kube-descheduler-operator
namespace: openshift-kube-descheduler-operator
spec:
channel: "4.9"
installPlanApproval: Automatic
name: cluster-kube-descheduler-operator
source: qe-app-registry
sourceNamespace: openshift-marketplace
startingCSV: clusterkubedescheduleroperator.4.9.0-202109071344

[cloud-user@preserve-olm-env jian]$ oc get sub
NAME PACKAGE SOURCE CHANNEL
cluster-kube-descheduler-operator cluster-kube-descheduler-operator qe-app-registry 4.9

[cloud-user@preserve-olm-env jian]$ oc get ip
NAME CSV APPROVAL APPROVED
install-4w9lm clusterkubedescheduleroperator.4.9.0-202109071344 Automatic true

[cloud-user@preserve-olm-env jian]$ oc get csv
NAME DISPLAY VERSION REPLACES PHASE
clusterkubedescheduleroperator.4.9.0-202109071344 Kube Descheduler Operator 4.9.0-202109071344 Succeeded
elasticsearch-operator.5.2.0-60 OpenShift Elasticsearch Operator 5.2.0-60 elasticsearch-operator.5.1.1-56 Succeeded

3, check the bundle content:
clusterkubedescheduleroperator.4.8.0-202108312109 {"apiVersion":"opera {"apiVersion":"operators.coreos.com/v1alpha1","kin registry-proxy.engineering.redhat.com/rh-osbs/openshift-ose- >=4.6.0 <4.8.0-202108312109 4.8.0-202108312109

clusterkubedescheduleroperator.4.9.0-202109071344 {"apiVersion":"opera {"apiVersion":"operators.coreos.com/v1alpha1","kin registry-proxy.engineering.redhat.com/rh-osbs/openshift-ose- >=4.6.0 <4.9.0-202109071344 4.9.0-202109071344
Version-Release number of selected component (if applicable):

[knarra@knarra ~]$ oc get clusterversion
NAME VERSION AVAILABLE PROGRESSING SINCE STATUS
version 4.9.0-0.nightly-2021-09-06-004132 True False 23h Cluster version is 4.9.0-0.nightly-2021-09-06-004132

Clusterkubedescheduleroperator.4.8.0-202108312109

How reproducible:
Install 4.8 cluster
Install 4.8 descheduler operator
Upgrade system to 4.9
Now edit descheduler sub & set channel to 4.9, starting csv to clusterkubedescheduleroperator.4.9.0-202109071344
spec:
channel: "4.9"
installPlanApproval: Automatic
name: cluster-kube-descheduler-operator
source: qe-app-registry
sourceNamespace: openshift-marketplace
startingCSV: clusterkubedescheduleroperator.4.9.0-202109071344

Actual Results:
Upgrade does not start at all

Expected Results:
Upgrade should work fine.