scheduler: resubmitting the same version of a system job results in a hanging "running" deployment in clusters with infeasible nodes #26886
Description
If you resubmit an unmodified system job in a cluster with infeasible nodes, the scheduler doesn't correctly handle it and instead of ignoring, it creates a hanging deployment in "running" state with an incorrect "Desired Total" number.