edit: TensorImpl: gave up original approach to use class partial

specialization because it is better handled by compilers

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

@@ -23,21 +23,7 @@ namespace Aidge {
 
 
 template <typename SRC_T, typename DST_T>
-    std::enable_if_t<
-              !(std::is_same<half_float::half, DST_T>::value
-                || std::is_same<half_float::half, SRC_T>::value)>
-thrust_copy(const SRC_T* srcData, DST_T* dstData, size_t size);
-
-void
-thrust_copy(const half_float::half* srcData, half_float::half* dstData, size_t size);
-
-template <typename DST_T>
-std::enable_if_t<!std::is_same<half_float::half, DST_T>::value>
-thrust_copy(const half_float::half* srcData, DST_T* dstData, size_t size);
-
-template <typename SRC_T>
-std::enable_if_t<!std::is_same<half_float::half, SRC_T>::value>
-thrust_copy(const SRC_T* srcData, half_float::half* dstData, size_t size);
+void thrust_copy(const SRC_T* srcData, DST_T* dstData, size_t size);
 
 /**
  * @brief Abstract class for the TensorImpl_cuda class template.