core: Change behavior of "application is alive" checks

Change some things in these "app is alive" checks:
- The dialog timeout is separated from the ping timeout, in order
  to show it again at a constant rate after dismissing, despite in
  flight pings. It still shows immediately after the first failed
  ping.
- As we want to tap further into is-alive logic, MetaWindow now
  made it a property, that other places in code can fetch and
  subscribe.
- Motion events trigger ping (as long as there was none other in
  flight for the same window), and are counted between ping and
  pong, in order to preemptively declare the window as not alive
  before there is trouble with event queues being overflown.

This results in a separate logic between "the application does
not respond" and "we are showing the close dialog" so that the
former may get triggered independently.

Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2122>

src/core/events.c

@@ -202,7 +202,7 @@ meta_display_handle_event (MetaDisplay        *display,
                            const ClutterEvent *event)
 {
   MetaBackend *backend = meta_get_backend ();
-  MetaWindow *window;
+  MetaWindow *window = NULL;
   gboolean bypass_clutter = FALSE;
   G_GNUC_UNUSED gboolean bypass_wayland = FALSE;
   MetaGestureTracker *gesture_tracker;
@@ -475,6 +475,10 @@ meta_display_handle_event (MetaDisplay        *display,
 #ifdef HAVE_WAYLAND
   if (compositor && !bypass_wayland)
     {
+      if (window && event->type == CLUTTER_MOTION &&
+          event->any.time != CLUTTER_CURRENT_TIME)
+        meta_window_check_alive_on_event (window, event->any.time);
+
       if (meta_wayland_compositor_handle_event (compositor, event))
         bypass_clutter = TRUE;
     }