Well, I can't say what happened exactly, but the problem server is now reported "9.0". I had done two things:
- Update-ClusterFunctionalLevel
- Reboot server.
But I'm not sure if #1 did anything. You see, the other two nodes in the cluster were already reporting 9.0 existing and I was able to upgrade VMs using "Update-VMVersion".
Oh well. At least all the VMs are being updated now.