e2e: move deleteState file to the AfterEach

Under the CPU manager and topology manager e2e tests possible the situation
when one of steps under the test will fail and it will not clean the CPU manager
state file. Move the deletion of the state file to `AfterEach` to guarantee that
the state file will be always removed from the node.

Signed-off-by: Artyom Lukianov <alukiano@redhat.com>

test/e2e_node/cpu_manager_test.go

@@ -627,7 +627,9 @@ func runCPUManagerTests(f *framework.Framework) {
 		ginkgo.By("wait for the deleted pod to be cleaned up from the state file")
 		waitForStateFileCleanedUp()
 		ginkgo.By("the deleted pod has already been deleted from the state file")
+	})
 
+	ginkgo.AfterEach(func() {
 		setOldKubeletConfig(f, oldCfg)
 	})
 }