fix: CNN test tool GPU readback with wgpuDevicePoll

Fixed buffer mapping callback mode mismatch causing Unknown status.
Changed from WaitAnyOnly+ProcessEvents to AllowProcessEvents+DevicePoll.

Readback now functional but CNN output incorrect (all white).
Issue isolated to tool-specific binding/uniform setup - CNNEffect
in demo works correctly.

Technical details:
- WGPUCallbackMode_WaitAnyOnly requires wgpuInstanceWaitAny
- Using wgpuInstanceProcessEvents with WaitAnyOnly never fires callback
- Fixed by using AllowProcessEvents mode + wgpuDevicePoll
- Removed debug output and platform warnings

Status: 36/36 tests pass, readback works, CNN shader issue remains.

handoff(Claude): CNN test tool readback fixed, output debugging needed

1 files changed, 7 insertions, 5 deletions

diff --git a/src/gpu/texture_readback.cc b/src/gpu/texture_readback.cc
index 3a690d3..0eb63d7 100644
--- a/src/gpu/texture_readback.cc
+++ b/src/gpu/texture_readback.cc
@@ -72,6 +72,9 @@ std::vector<uint8_t> read_texture_pixels(
   wgpuCommandBufferRelease(commands);
   wgpuCommandEncoderRelease(encoder);
 
+  // Wait for copy to complete before mapping
+  wgpuDevicePoll(device, true, nullptr);
+
   // Map buffer for reading (API differs between Win32 and native)
 #if defined(DEMO_CROSS_COMPILE_WIN32)
   // Win32: Old callback API
@@ -95,7 +98,7 @@ std::vector<uint8_t> read_texture_pixels(
     state->done = true;
   };
   WGPUBufferMapCallbackInfo map_info = {};
-  map_info.mode = WGPUCallbackMode_WaitAnyOnly;
+  map_info.mode = WGPUCallbackMode_AllowProcessEvents;  // Fire during ProcessEvents
   map_info.callback = map_cb;
   map_info.userdata1 = &map_state;
   wgpuBufferMapAsync(staging, WGPUMapMode_Read, 0, buffer_size, map_info);
@@ -106,14 +109,13 @@ std::vector<uint8_t> read_texture_pixels(
 #if defined(__EMSCRIPTEN__)
     emscripten_sleep(10);
 #else
-    wgpuInstanceProcessEvents(instance);
+    wgpuDevicePoll(device, true, nullptr);
 #endif
   }
 
-  if (map_state.status != WGPUMapAsyncStatus_Success) {
-    fprintf(stderr, "Buffer mapping failed: %d\n", map_state.status);
+  if (!map_state.done || map_state.status != WGPUMapAsyncStatus_Success) {
     wgpuBufferRelease(staging);
-    return pixels; // Return empty
+    return pixels; // Return empty on timeout or failure
   }
 
   // Copy data from mapped buffer (handle row padding)