Resolve "Arithmetic Operator optimization"

Closes #5

• Add
• And
• BitShift
• Div
• Equal
• Mul
• Pow
• Sub

Here is the comparison of performances for Add operator in percents (%). It is representative of perfomance gains for other elementwise operators.

Default size of both inputs was [64,64,64,64]

I also changed Catch2 version from 3.0.1 to 3.7.1

include/aidge/backend/cpu/operator/AndImpl_kernels.hpp

@@ -144,21 +144,20 @@ void EqualImpl_cpu_forward_kernel(std::vector<std::size_t> dims0,
     }
 }
 
-} // namespace Aidge
-
 // Kernels registration to implementation entry point
 REGISTRAR(AndImpl_cpu,
     {DataType::Float32},
-    {ProdConso::inPlaceModel, Aidge::EqualImpl_cpu_forward_kernel<float, float, float>, nullptr});
+    {ProdConso::inPlaceModel, Aidge::EqualImpl_cpu_forward_kernel<float, float>, nullptr});
 REGISTRAR(AndImpl_cpu,
     {DataType::Float64},
-    {ProdConso::inPlaceModel, Aidge::EqualImpl_cpu_forward_kernel<double, double, double>, nullptr});
+    {ProdConso::inPlaceModel, Aidge::EqualImpl_cpu_forward_kernel<double, double>, nullptr});
 REGISTRAR(AndImpl_cpu,
     {DataType::Int32},
-    {ProdConso::inPlaceModel, Aidge::EqualImpl_cpu_forward_kernel<std::int32_t, std::int32_t, std::int32_t>, nullptr});
+    {ProdConso::inPlaceModel, Aidge::EqualImpl_cpu_forward_kernel<std::int32_t, std::int32_t>, nullptr});
 REGISTRAR(AndImpl_cpu,
     {DataType::Int64},
-    {ProdConso::inPlaceModel, Aidge::EqualImpl_cpu_forward_kernel<std::int64_t, std::int64_t, std::int64_t>, nullptr});
+    {ProdConso::inPlaceModel, Aidge::EqualImpl_cpu_forward_kernel<std::int64_t, std::int64_t>, nullptr});
+
 }  // namespace Aidge
 
 #endif /* AIDGE_CPU_OPERATOR_ANDIMPL_KERNELS_H_ */