test(gfx): 新增 custom-pipeline compute shader 测试场景与自动化用例

- cases/GFX/compute-demo: compute-demo.scene + raytracingByCompute.ts（RaytracingWeekend 自定义管线，compute pass 写 storage image 后全屏 blit 输出）
- resources/raytracing-compute: rt-compute（compute shader）+ rt-swizzle（输出）effect/material，保留 meta 以维持 mtl->effect uuid 引用
- auto-test-case/dynamic/GFX/compute-demo.test.ts: 截图式动态测试，import 管线模块确保 setCustomPipeline 注册
- auto-test-config.json: sceneList 注册 compute-demo
- 注意：3.8.4 运行需在项目设置将 Render Pipeline 切换为 custom-pipeline

Author: vanitynie

assets/auto-test-case/dynamic/GFX/compute-demo.test.ts

@@ -0,0 +1,19 @@
+//@ts-ignore
+import { director } from 'cc';
+//@ts-ignore
+import { runScene, testCase, testClass } from 'db://automation-framework/runtime/test-framework.mjs';
+import { screenshot_custom } from '../common/utils';
+// ensure the compute pipeline module is loaded/registered before the scene renders
+import '../../../cases/GFX/compute-demo/raytracingByCompute';
+
+@runScene('compute-demo')
+@testClass('ComputeDemo')
+export class ComputeDemo {
+    // give the compute pipeline a few frames to create storage/dispatch resources
+    _delay = 5;
+
+    @testCase
+    async index() {
+        await screenshot_custom(this._delay);
+    }
+}

assets/auto-test-case/dynamic/GFX/compute-demo.test.ts.meta

@@ -0,0 +1,9 @@
+{
+  "ver": "4.0.24",
+  "importer": "typescript",
+  "imported": true,
+  "uuid": "67fcc6d9-9d47-4afa-ad24-615700d5df98",
+  "files": [],
+  "subMetas": {},
+  "userData": {}
+}

assets/auto-test-config.json

@@ -218,6 +218,7 @@
         "ui-opacity",
         "SpineAttachment",
         "SpineBlend",
-        "SpinePhysics"
+        "SpinePhysics",
+        "compute-demo"
     ]
 }

assets/cases/GFX/compute-demo.meta

@@ -0,0 +1,9 @@
+{
+  "ver": "1.2.0",
+  "importer": "directory",
+  "imported": true,
+  "uuid": "1c1a1db3-316b-49b0-811a-054e2d66cd12",
+  "files": [],
+  "subMetas": {},
+  "userData": {}
+}

assets/cases/GFX/compute-demo/README.md

@@ -0,0 +1,45 @@
+# compute-demo (Custom Pipeline / Compute Shader)
+
+Mirror of Cocos official tutorial-005-compute (raytracing by compute) from
+cocos-example-custom-pipeline, adapted as an automation test case.
+
+## 运行前提（重要）
+
+1. **后端（Backend）**：浏览器预览必须先开启 **WebGPU** —— 使用支持 WebGPU 的
+   Chrome / Edge 等浏览器（较新版本默认开启；旧版本需在 chrome://flags 打开
+   WebGPU 相关开关）。桌面预览可用 Vulkan / Metal。
+2. **渲染管线（图形设置）**：项目设置 -> 功能裁剪（Feature Cropping）-> 图形 /
+   渲染管线选择**新的渲染管线（custom-pipeline）**，并在管线名称（Custom Pipeline
+   Name）中填写 **RaytracingWeekend** —— 必须与 raytracingByCompute.ts 中
+   rendering.setCustomPipeline('RaytracingWeekend', ...) 的注册名一致，否则管线不会生效。
+
+## 文件说明
+
+- raytracingByCompute.ts - PipelineBuilder + rendering.setCustomPipeline('RaytracingWeekend', ...)
+  (compute pass writes to a storage texture, a fullscreen quad blits it to the swapchain)
+- compute-demo.scene - camera + main light only, compute output fills the window
+- ../pipeline-data.ts - window/framebuffer helper imported by the pipeline script
+- assets/resources/raytracing-compute/ - rt-compute (compute shader) and
+  rt-swizzle (blit) effects + materials, loaded at runtime via resources.load
+
+## QA 步骤
+
+1. 按"运行前提"完成配置：浏览器预览开启 WebGPU；图形设置选新渲染管线，
+   管线名称填 RaytracingWeekend。
+2. 打开场景 compute-demo.scene。手动预览时，把 raytracingByCompute.ts
+   （组件 pipeline_005_raytracing_in_1_weekend）挂到任意节点上，
+   否则管线不会被注册。
+3. 预览或跑自动化验证 compute 输出。
+
+## 自动化
+
+- Registered in assets/auto-test-config.json under sceneList as compute-demo.
+- Test: assets/auto-test-case/dynamic/GFX/compute-demo.test.ts (screenshot based).
+
+## 维护
+
+- Edit scene content in rt-compute.effect (spheres uniform, imageStore).
+- addDispatch(width/8, height/4, 1) must match layout(local_size_x = 8, local_size_y = 4).
+- The .effect.meta/.mtl.meta files are kept on purpose: .mtl binds the
+  effect by uuid. If you delete them, re-assign the effect on both materials
+  after Creator regenerates the metas.

assets/cases/GFX/compute-demo/README.md.meta

@@ -0,0 +1,11 @@
+{
+  "ver": "1.0.1",
+  "importer": "text",
+  "imported": true,
+  "uuid": "8bff7cee-33e4-4903-8f38-87f2e4c243ca",
+  "files": [
+    ".json"
+  ],
+  "subMetas": {},
+  "userData": {}
+}