Use RenderStartup in custom_post_processing example (#19886)

# Objective

- This example uses a FromWorld impl to initialize a resource on startup
- #19887

## Solution

- Use RenderStartup instead

## Testing

- The example still works as expected

examples/shader/custom_post_processing.rs

@@ -27,7 +27,7 @@ use bevy::{
         },
         renderer::{RenderContext, RenderDevice},
         view::ViewTarget,
-        RenderApp,
+        RenderApp, RenderStartup,
     },
 };
 
@@ -66,6 +66,10 @@ impl Plugin for PostProcessPlugin {
             return;
         };
 
+        // RenderStartup runs once on startup after all plugins are built
+        // It is useful to initialize data that will only live in the RenderApp
+        render_app.add_systems(RenderStartup, setup_pipeline);
+
         render_app
             // Bevy's renderer uses a render graph which is a collection of nodes in a directed acyclic graph.
             // It currently runs on each view/camera and executes each node in the specified order.
@@ -97,17 +101,6 @@ impl Plugin for PostProcessPlugin {
                 ),
             );
     }
-
-    fn finish(&self, app: &mut App) {
-        // We need to get the render app from the main app
-        let Some(render_app) = app.get_sub_app_mut(RenderApp) else {
-            return;
-        };
-
-        render_app
-            // Initialize the pipeline
-            .init_resource::<PostProcessPipeline>();
-    }
 }
 
 #[derive(Debug, Hash, PartialEq, Eq, Clone, RenderLabel)]
@@ -233,10 +226,13 @@ struct PostProcessPipeline {
     pipeline_id: CachedRenderPipelineId,
 }
 
-impl FromWorld for PostProcessPipeline {
-    fn from_world(world: &mut World) -> Self {
-        let render_device = world.resource::<RenderDevice>();
-
+fn setup_pipeline(
+    mut commands: Commands,
+    render_device: Res<RenderDevice>,
+    asset_server: Res<AssetServer>,
+    fullscreen_shader: Res<FullscreenShader>,
+    pipeline_cache: Res<PipelineCache>,
+) {
     // We need to define the bind group layout used for our pipeline
     let layout = render_device.create_bind_group_layout(
         "post_process_bind_group_layout",
@@ -253,17 +249,14 @@ impl FromWorld for PostProcessPipeline {
             ),
         ),
     );
-
     // We can create the sampler here since it won't change at runtime and doesn't depend on the view
     let sampler = render_device.create_sampler(&SamplerDescriptor::default());
 
     // Get the shader handle
-        let shader = world.load_asset(SHADER_ASSET_PATH);
+    let shader = asset_server.load(SHADER_ASSET_PATH);
     // This will setup a fullscreen triangle for the vertex state.
-        let vertex_state = world.resource::<FullscreenShader>().to_vertex_state();
-
-        let pipeline_id = world
-            .resource_mut::<PipelineCache>()
+    let vertex_state = fullscreen_shader.to_vertex_state();
+    let pipeline_id = pipeline_cache
         // This will add the pipeline to the cache and queue its creation
         .queue_render_pipeline(RenderPipelineDescriptor {
             label: Some("post_process_pipeline".into()),
@@ -289,13 +282,11 @@ impl FromWorld for PostProcessPipeline {
             push_constant_ranges: vec![],
             zero_initialize_workgroup_memory: false,
         });
-
-        Self {
+    commands.insert_resource(PostProcessPipeline {
         layout,
         sampler,
         pipeline_id,
-        }
-    }
+    });
 }
 
 // This is the component that will get passed to the shader