Merge branch 'unified_params' into 'main'

Changes following MR

See merge request !1

include/aidge/backend/cuda/data/TensorImpl.hpp

@@ -58,7 +58,11 @@ class TensorImpl_cuda : public TensorImpl, public TensorImpl_cuda_  {
     void *rawPtr() override {
         lazyInit(reinterpret_cast<void**>(&mData));
         return mData;
-    };
+    }
+
+    void* getRaw(std::size_t idx) {
+        return static_cast<void*>(static_cast<T*>(rawPtr()) + idx);
+    }
 
     const cudnnTensorDescriptor_t& getCudnnTensorDesc() const override {
         if (mCudnnTensor == nullptr) {

src/operator/ConvImpl.cpp

@@ -80,9 +80,9 @@ void Aidge::ConvImpl_cuda<DIM>::forward() {
 
     // Lazy-initialize CuDNN convolution descriptor
     if (mConvDesc == nullptr) {
-        const std::vector<int> strides(mOp.template get<ConvParam::StrideDims>().begin(), mOp.template get<ConvParam::StrideDims>().end());
+        const std::vector<int> strides(mOp.template getAttr<ConvAttr::StrideDims>().begin(), mOp.template getAttr<ConvAttr::StrideDims>().end());
         const std::vector<int> paddings(DIM, 0);
-        const std::vector<int> upscales(mOp.template get<ConvParam::DilationDims>().begin(), mOp.template get<ConvParam::DilationDims>().end());
+        const std::vector<int> upscales(mOp.template getAttr<ConvAttr::DilationDims>().begin(), mOp.template getAttr<ConvAttr::DilationDims>().end());
 
         CHECK_CUDNN_STATUS(cudnnCreateConvolutionDescriptor(&mConvDesc));
         CHECK_CUDNN_STATUS(