feat: GPU procedural Phase 4 - texture composition

Multi-input composite shaders with sampler support.
- Dynamic bind group layouts (N input textures + 1 sampler)
- dispatch_composite() for multi-input compute dispatch
- create_gpu_composite_texture() API
- gen_blend.wgsl and gen_mask.wgsl shaders

Guarded with #if !defined(STRIP_GPU_COMPOSITE) for easy removal.

Tests:
- Blend two noise textures
- Mask noise with grid
- Multi-stage composite (composite of composites)

Size: ~830 bytes (2 shaders + dispatch logic)

handoff(Claude): GPU procedural Phase 4 complete

1 files changed, 29 insertions, 0 deletions

diff --git a/assets/final/shaders/compute/gen_blend.wgsl b/assets/final/shaders/compute/gen_blend.wgsl
new file mode 100644
index 0000000..9fc9e1e
--- /dev/null
+++ b/assets/final/shaders/compute/gen_blend.wgsl
@@ -0,0 +1,29 @@
+// This file is part of the 64k demo project.
+// GPU composite shader: Blend two textures.
+
+struct BlendParams {
+  width: u32,
+  height: u32,
+  blend_factor: f32,
+  _pad0: f32,
+}
+
+@group(0) @binding(0) var output_tex: texture_storage_2d<rgba8unorm, write>;
+@group(0) @binding(1) var<uniform> params: BlendParams;
+@group(0) @binding(2) var input_a: texture_2d<f32>;
+@group(0) @binding(3) var input_b: texture_2d<f32>;
+@group(0) @binding(4) var tex_sampler: sampler;
+
+@compute @workgroup_size(8, 8, 1)
+fn main(@builtin(global_invocation_id) id: vec3<u32>) {
+  if (id.x >= params.width || id.y >= params.height) { return; }
+
+  let uv = vec2<f32>(f32(id.x) / f32(params.width),
+                      f32(id.y) / f32(params.height));
+
+  let color_a = textureSampleLevel(input_a, tex_sampler, uv, 0.0);
+  let color_b = textureSampleLevel(input_b, tex_sampler, uv, 0.0);
+  let blended = mix(color_a, color_b, params.blend_factor);
+
+  textureStore(output_tex, id.xy, blended);
+}