in our CI, we noticed that a k8s pod reports the wrong imageID in case the target image is identical to a previously pulled one

OCP 4.21.0

100% in our ci, will require some effort to craft a local reproducer but I imagine for identical manifests lists/images this should trigger

1. create identical manifest list with images on two different registries
2. pull from registry A
3. pull from registry B

imageID points at a different non-user-specified registry

imageID deduced from the hash/registry specified by user    

with no mirrors configured, this is the end result

"image": "registry:5000/kubevirt/alpine-container-disk-demo:devel",
"imageID": "quay.io/kubevirt/alpine-container-disk-demo@sha256:081ab0c41a37499672990f612488fe0cded25589b60593473604df5d758afe8b",

this bites us in kubevirt since we craft the exact hash of VM images when migrating, in case the tag has changed since. in this case, the hash/registry is wrong