During the pod evacuation during an update process in OCP 4.x, the scheduler should select the newest nodes

1. Proposed title of this feature request
During the pod evacuation in a update process in OCP 4.x, the scheduler should select the newest nodes
2. What is the nature and description of the request?
In an update process, you have a lot of reboots (lost of time) because pod evacuation. The scheduler assign the pods randomly. If you have a huge cluster (100 nodes) you probably spent a lot of time during the pod evacuation. It would be a good idea that the scheduler try to evacuate the pods to new nodes to reduce the upgrade time.
3. Why does the customer need this? (List the business requirements here)
This customer has a small update window, a lot of clusters and a huge ones (more than 100 nodes). They need to reduce their update time as much as possible in order to update all their infrastructure to the minimun amount of time.
4. List any affected packages or components.
Scheduler