CUDA: fix round issue, under MSVC the CPU and GPU agree (#30229)

ggml/src/ggml-cuda/unary.cu

@@ -107,7 +107,7 @@ static __device__ __forceinline__ float op_ceil(float x) {
 }
 
 static __device__ __forceinline__ float op_round(float x) {
-    return round(x);
+    return roundf(x);
 }
 
 static __device__ __forceinline__ float op_trunc(float x) {