Add TensorFusion hypervisor Pod into Karpenter DaemonSet-like pod

Karpenter provisioning and bin-packing judges Pod's type at the beginning, when it's DaemonSet-like pod, it's not reschedulable and provisionable. But Hypervisor Pod is owned by GPUNode custom resource, even it's started on each Node, it won't be recognized as DaemonSet-like Pod, causing Karpenter integration bugs.