Defining common operators for std::integral_constant

You can run this code

#include <type_traits>

int main()
{
  std::integral_constant<int, 3> a;
  std::integral_constant<int, 2> b;
  auto c = a + b;

  static_assert(std::is_same<decltype(c), int>::value);
}

to verify that c type is int.

As the compiler has all the required information, on my side I prefer to get a std::integral_constant<int, 5> type, this guarantee compile-time computations and reduced size storage. That's the reason why, I often use stuff like:

#include <type_traits>

template <typename T, T A, T B>
std::integral_constant<T, A + B> operator+(
    const std::integral_constant<T, A>& a,
    const std::integral_constant<T, B>& b) noexcept
{
  return decltype(a + b)();
}

int main()
{
  std::integral_constant<int, 3> a;
  std::integral_constant<int, 2> b;
  auto c = a + b;
  static_assert(std::is_same<decltype(c), std::integral_constant<int, 5>>::value);
}

in my code.

I have two questions:

• Were there any discussions to define such operator overloading in the C++ std lib?
• Are there any risks in defining such overloads in our code (interference with the standard library, side effect)?