feat(render3d): compute and indirect draws on Vulkan, and tree layers culled on the GPU (opt-in)

- Device: multiDrawIndirect, drawIndirectFirstInstance and drawIndirectCount where present.
- Buffers carry storage and indirect usage; a GPU-owned buffer is never swapped under a draw.
- Compute programs from shaders/compute.list (binding 0 parameters, 1.. storage buffers),
  built by `ludic-dev shaders`; gpu_compute / gpu_dispatch / gpu_draw_mesh_indirect in gpu.ludic.
- R3D_VK_PROBE=1: a dispatch read back (OK on the RTX 3070 Ti).
- scatter_cull.comp: a tree layer's frustum test and LOD split on the GPU, with the lit, prepass,
  impostor and shadow-LOD draws reading its records. Behind R3D_GPU_CULL=1 and off by default:
  at the camp it is slower (43.0 fps against 53.3), because the frame's cost is per-draw
  descriptor sets and it adds empty-level draws. Validation-clean; OpenGL frames unchanged.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

packages/ludic.render3d/shaders/probe.comp

@@ -0,0 +1,10 @@
+// probe.comp - the compute path's own check (R3D_VK_PROBE=1): every float in the buffer times
+// the parameter, so a readback proves the dispatch, the bindings and the GPU-owned buffer.
+layout(local_size_x = 64) in;
+layout(set = 0, binding = 0) uniform Params { uint count; float mul; } pr;
+layout(set = 0, binding = 1) buffer Data { float v[]; } data;
+void main() {
+  uint i = gl_GlobalInvocationID.x;
+  if (i >= pr.count) { return; }
+  data.v[i] = data.v[i] * pr.mul;
+}