UPSTREAM: <carry>: Extend default unit test timeout to 180s

The following packages have tests that exceed the default 120s
timeout:

k8s.io/kubernetes/pkg/kubelet/volumemanager/reconciler

 - The tests in this package collectively take longer than 120s.

k8s.io/kubernetes/pkg/volume/csi

 - One of the unit tests has to wait 2 minutes for a timeout to
   validate its failure condition.

hack/make-rules/test.sh

@@ -59,7 +59,11 @@ kube::test::find_dirs() {
   )
 }
 
-KUBE_TIMEOUT=${KUBE_TIMEOUT:--timeout=120s}
+# The default timeout of 120s is not sufficient for the following packages
+# whose successful execution takes longer than that.
+# - k8s.io/kubernetes/pkg/kubelet/volumemanager/reconciler
+# - k8s.io/kubernetes/pkg/volume/csi
+KUBE_TIMEOUT=${KUBE_TIMEOUT:--timeout=180s}
 KUBE_COVER=${KUBE_COVER:-n} # set to 'y' to enable coverage collection
 KUBE_COVERMODE=${KUBE_COVERMODE:-atomic}
 # The directory to save test coverage reports to, if generating them. If unset,

pkg/volume/csi/csi_attacher_test.go

@@ -949,6 +949,13 @@ func TestAttacherDetach(t *testing.T) {
 				return false, nil, nil
 			},
 		},
+		// The watcher error triggered by this test case will result in the
+		// Detach() call on L984 timing out after 2 minutes since an error
+		// condition is logged but not returned. Thus this test will only pass if
+		// the default value for KUBE_TEST_TIMEOUT in hack/make-rules/test.sh is
+		// greater than the default of 120s. It's not clear why this test was
+		// merged at all given that a watcher error is only logged and cannot be
+		// observed by the test.
 		{
 			name:         "API watch error happen",
 			volID:        "vol-005",