Specifically, leaving the corporate political dynamics aside, we move the workloads to the deployments that are up to standard and then archive+drop the old ones. Very simple.
A more cynical man than I would say that if you need to recreate all your workloads on a new cluster to bring them up to standard, "you seriously fucked up as DevOps / cloud admin / whatever"
Heh, they aren't ""our"" workloads, before we get there, and not until they get deployed correctly for the first time ever at which point then and only then are they ""our"" workloads.