Description of problem:
while upgrade redeploying of monitoring daemons fail saying registry login required when registry login was done manually during deployment.

The failure is awkward as podman login was performed initially without which cluster would not have been able to configure.

Version-Release number of selected component (if applicable):
16.2.8-45.el8cp

How reproducible:
Thrice in different envs

Steps to Reproduce:
1. Configure ceph cluster using cephadm.
have two different container registries for monitoring stack daemons and ceph.
Monitoring stack image registry needs to have authentication and do podman login manually using `podman login <registry>`
2. Upgrade cluster to another build using ceph orch upgrade.

Actual results:
Upgrade fails saying prometheus daemon redeploy failed due to invalid credentials .

Expected results:
registry login must not be asked again as it was performed during initial cluster deployment and when there had not been any change in the credentials.


Additional info:
upgrade failure was not repeated after using cephadm registry-login on that particular node.