Goal:

Watcher default value for max_workers parameter in watcher_workflow_engines.taskflow section is set to the númber of CPUs in the server running the watcher-applier pod. When running the control plane in big controllers, that may be a too high when running workload_stabilization and node_resource_consolidation strategies and would end up running too many actions (vm migrations in most cases) in parallel.

To protect shared network bandwidth, we need to cap parallel actions. Setting a lower, static defensive value for max_workers prevents high-spec controllers from triggering excessive concurrent actions. The admin may still override it by using customServiceConfig.