What would you like to be added:
Currently, kubernetes scheduler supports various scheduling primitives, such as pod affinity/anti-affinity, pod disruption budget, pod topology spread constraints and so on.

Why is this needed:
some scheduling primitives are provided by prefered logical instead of required which may in the lifecycle of the workload, some break variation may happen because pod priority preemption or node pressure eviction.

I think we need to collect the spec and status of the workload pod distribution or the dummy scheduler attribute for every workload and do some alarms when needed.
Describe the solution you'd like
NA. I have not consider a solution about this. Will give it a try in the later weekends.
Additional context
NA