Treat first-seen, non-pending pods as updates

Kubelet doesn't perform checkpointing and loses all its internal states after
restarts. It'd then mistaken pods from the api server as new pods and attempt
to go through the admission process. This may result in pods being rejected
even though they are running on the node (e.g., out of disk situation). This
change adds a condition to check whether the pod was seen before and categorize
such pods as updates. The change also removes freeze/unfreeze mechanism used to
work around such cases, since it is no longer needed and it stopped working
correctly ever since we switched to incremental updates.

pkg/kubelet/kubelet_test.go

@@ -2482,7 +2482,6 @@ func updateDiskSpacePolicy(kubelet *Kubelet, mockCadvisor *cadvisor.Mock, rootCa
 	if err != nil {
 		return err
 	}
-	diskSpaceManager.Unfreeze()
 	kubelet.diskSpaceManager = diskSpaceManager
 	return nil
 }
@@ -3612,6 +3611,15 @@ func TestRegisterExistingNodeWithApiserver(t *testing.T) {
 		DockerVersion:      "1.5.0",
 	}
 	mockCadvisor.On("VersionInfo").Return(versionInfo, nil)
+	mockCadvisor.On("DockerImagesFsInfo").Return(cadvisorapiv2.FsInfo{
+		Usage:     400 * mb,
+		Capacity:  1000 * mb,
+		Available: 600 * mb,
+	}, nil)
+	mockCadvisor.On("RootFsInfo").Return(cadvisorapiv2.FsInfo{
+		Usage:    9 * mb,
+		Capacity: 10 * mb,
+	}, nil)
 
 	done := make(chan struct{})
 	go func() {