Optimize render graph and add GPU profiling

Amp-Thread-ID: https://ampcode.com/threads/T-01a01380-b478-77d0-84a0-102880a5c5ae
Co-authored-by: Heaust Azure <heaust.azure@gmail.com>

docs/guide/core.md

@@ -39,10 +39,25 @@ const id = await core.allocateObject("application.values");
 values.row(id).set([1, 2, 3, 4]);
 ```
 
-Graph frontends serialize plain data to `(yawn-graph 1 ...)`. Named `after` edges preserve DAG fan-out; Rust sorts passes, detects cycles, culls unused declarations, plans compatible transient lifetimes, and allocates the active loadout.
+Graph frontends serialize plain data to `(yawn-graph 1 ...)`. Named `after` edges preserve DAG fan-out; Rust sorts passes, detects cycles, culls unused declarations, aliases compatible transient lifetimes, merges compatible render passes into bundles, and allocates the active loadout. Unchanged persistent textures survive loadout rebuilds.
 
 The `Scene` addon is one such frontend. It is replaceable and has no privileged core API.
 
+## Profile physical GPU passes
+
+```ts
+const stop = core.onProfile((frame) => {
+  console.table(frame.passes); // name + GPU milliseconds
+});
+const supported = await core.setProfiler(true);
+
+// Later:
+await core.setProfiler(false);
+stop();
+```
+
+`new YawnCore(canvas, { debug: true })` enables the same timestamp-query mode at startup. Timings describe the physical compute and render passes produced by graph compilation, so 138 compatible draws appear as one bundled forward pass. The fullscreen playground’s **Profile** button shows the stream in a toggleable sidebar.
+
 <Playground example="core" />
 
 <script setup>

docs/guide/materials.md

@@ -22,17 +22,15 @@ mesh.material = paint;
 
 ```ts
 const albedo = new Texture(scene, {
-  source: "/textures/paint.ktx2",
-  size: [2048, 2048, 1],
+  source: "/textures/paint.png",
   format: "rgba8unorm-srgb",
-  usage: ["sampled", "copyDst"],
 });
 await albedo.ready;
 
 const textured = new PBRMaterial(scene, { baseColorTexture: albedo });
 ```
 
-Creating or removing a `Texture` rebuilds the single graph loadout so the GPU resource is allocated up front. The source pointer stays on the handle for an importer or application uploader; core never owns image-loading policy.
+Creating or removing a `Texture` rebuilds the single graph loadout so the GPU resource is allocated up front. The addon decodes URLs outside core, then transfers an `ImageBitmap` to the render worker; compatible loadout rebuilds reuse the allocated GPU texture instead of uploading it again.
 
 ## Custom WGSL