Ensure spawning related entities in an OnAdd observer downstream of a World::spawn in a Command does not cause a crash (#18545)

# Objective

fixes #18452.

## Solution

Spawning used to flush commands only, but those commands can reserve
entities. Now, spawning flushes everything, including reserved entities.
I checked, and this was the only place where `flush_commands` is used
instead of `flush` by mistake.

## Testing

I simplified the MRE from #18452 into its own test, which fails on main,
but passes on this branch.

Author: ElliottjPierce

crates/bevy_ecs/src/observer/mod.rs

@@ -1648,6 +1648,23 @@ mod tests {
         assert_eq!(vec!["event", "event"], world.resource::<Order>().0);
     }
 
+    // Originally for https://github.com/bevyengine/bevy/issues/18452
+    #[test]
+    fn observer_modifies_relationship() {
+        fn on_add(trigger: Trigger<OnAdd, A>, mut commands: Commands) {
+            commands
+                .entity(trigger.target())
+                .with_related::<crate::hierarchy::ChildOf>(|rsc| {
+                    rsc.spawn_empty();
+                });
+        }
+
+        let mut world = World::new();
+        world.add_observer(on_add);
+        world.spawn(A);
+        world.flush();
+    }
+
     // Regression test for https://github.com/bevyengine/bevy/issues/14467
     // Fails prior to https://github.com/bevyengine/bevy/pull/15398
     #[test]

crates/bevy_ecs/src/world/mod.rs

@@ -1174,7 +1174,7 @@ impl World {
 
         // SAFETY: command_queue is not referenced anywhere else
         if !unsafe { self.command_queue.is_empty() } {
-            self.flush_commands();
+            self.flush();
             entity_location = self
                 .entities()
                 .get(entity)