fix : impossible conversion from float ** to void **

src/operator/AddImpl.cpp

@@ -195,7 +195,7 @@ void Aidge::AddImpl_cuda::backward_(const Tensor& outputGrad, const std::vector<
                                tensorDesc,
                                &workspaceSize));
 
-            float *d_workspace;
+            void *d_workspace;
             CHECK_CUDA_STATUS(cudaMalloc(&d_workspace, workspaceSize));
 
             CHECK_CUDNN_STATUS(cudnnReduceTensor(CudaContext::cudnnHandle(),

src/operator/SubImpl.cpp

@@ -201,7 +201,7 @@ void Aidge::SubImpl_cuda::backward_(const Tensor& outputGrad, const std::vector<
                                tensorDesc,
                                &workspaceSize));
 
-            float *d_workspace;
+            void *d_workspace;
             CHECK_CUDA_STATUS(cudaMalloc(&d_workspace, workspaceSize));
 
             CHECK_CUDNN_STATUS(cudnnReduceTensor(CudaContext::cudnnHandle(),
@@ -220,4 +220,4 @@ void Aidge::SubImpl_cuda::backward_(const Tensor& outputGrad, const std::vector<
             CHECK_CUDNN_STATUS(cudnnDestroyTensorDescriptor(tensorDesc));
         }
     }
-}
\ No newline at end of file
+}