[CUDA] synch properly waits for all tasks to finish and clear (#2303)

* cuda synch properly waits for all tasks to finish and clear

* fix copy

mlx/backend/cuda/device.h

@@ -123,6 +123,9 @@ class CommandEncoder {
     return has_gpu_work_;
   }
 
+  // Wait until kernels and completion handlers are finished
+  void synchronize();
+
  private:
   Device& device_;
   DeviceStream& stream_;