Make storage unmount less strict

There are cases where the container storage unmount has been already
(partially) done. This would cause `StopContainer()`  in
`server/container_stop.go:76` fail and therefore make containers get
stuck in recreation, making their pods stuck in `NotReady`.

We now double check the two c/stroage errors `ErrContainerUnknown` and
`ErrLayerUnknown`

Somehow related to:
containers/podman#11207 (comment)

Signed-off-by: Sascha Grunert <sgrunert@redhat.com>

internal/storage/runtime.go

@@ -455,11 +455,22 @@ func (r *runtimeService) StopContainer(idOrName string) error {
 	if err != nil {
 		return err
 	}
-	_, err = r.storageImageServer.GetStore().Unmount(container.ID, false)
-	if err != nil {
-		logrus.Debugf("Failed to unmount container %q: %v", container.ID, err)
+
+	if _, err := r.storageImageServer.GetStore().Unmount(container.ID, true); err != nil {
+		if errors.Is(err, storage.ErrContainerUnknown) {
+			logrus.Infof("Storage for container %s already removed", container.ID)
+			return nil
+		}
+
+		if errors.Is(err, storage.ErrLayerUnknown) {
+			logrus.Infof("Layer for container %s not known", container.ID)
+			return nil
+		}
+
+		logrus.Warnf("Failed to unmount container %q: %v", container.ID, err)
 		return err
 	}
+
 	logrus.Debugf("Unmounted container %q", container.ID)
 	return nil
 }