feat: Implement CUDA support for VoxRaytracer, add CUDA tests for voxel image operations, and update CMake to enable CUDA compilation.

src/Math/testing/VoxRaytracerTest.cpp

@@ -48,6 +48,11 @@ int Vector4f0(Vector4f c) {
 
 typedef VoxRaytracer Raytracer;
 
+struct TestVoxel {
+  float Value;
+  int Count;
+};
+
 int main() {
   BEGIN_TESTING(Math VoxRaytracer);
 
@@ -132,5 +137,41 @@ int main() {
     ray.PrintSelf(std::cout);
   }
 
+#ifdef USE_CUDA
+  {
+    std::cout << "\n--- Testing CUDA Raytracer Accumulator ---\n";
+
+    VoxImage<TestVoxel> img_cuda(Vector3i(4, 4, 4));
+    img_cuda.SetSpacing(Vector3f(2, 2, 2));
+    img_cuda.SetPosition(Vector3f(-4, -4, -4));
+
+    Raytracer ray(img_cuda);
+
+    HLine3f line1;
+    line1.origin << -3, -3, -3, 1;
+    line1.direction << 1, 1, 1, 0;
+
+    HLine3f line2;
+    line2.origin << -3, -3, 1, 1;
+    line2.direction << 1, 1, -1, 0;
+
+    HLine3f lines[2] = {line1, line2};
+    // Execute CUDA kernel wrapper over target VoxImage mapped internally into
+    // VRAM
+    ray.AccumulateLinesCUDA(lines, 2, img_cuda);
+
+    // Validate device synchronization returned data correctly pulling back to
+    // host
+    TEST1(img_cuda.Data().GetDevice() !=
+          MemoryDevice::RAM); // Confirms VRAM executed
+
+    // Pull down checking values
+    float l_val = img_cuda[img_cuda.Find(Vector4f(-3, -3, -3, 1))].Value;
+    std::cout << "Accumulated Voxel test trace point length returned: " << l_val
+              << "\n";
+    TEST1(l_val > 0.1f);
+  }
+#endif
+
   END_TESTING
 }