[CUDA] Reduce use of managed memory (#2725)

* Use async cuda malloc managed with cuda 13

* add pool threshold

* refactor for regular cuda malloc

* load eval gpu for cuda

* remove use of cuda pool, use cuda free async

* fix

* fix

* fix

* fix

* fix + comment

mlx/backend/cuda/kernel_utils.cuh

@@ -9,6 +9,7 @@
 #include <type_traits>
 
 #include "mlx/array.h"
+#include "mlx/backend/cuda/allocator.h"
 #include "mlx/backend/cuda/device/utils.cuh"
 
 #include <cuda.h>