`std::to_string` for extended floating-point types

[eisenwave]

std::to_string(x) is currently defined in terms of std::format("{}", x), which is defined in terms of std::to_chars. There is no technical reason why std::to_chars but not std::to_string should support the printing of extended floating-point types.

The solution would be to replace the overloads for float, double, and long double with a single floating-point-type overload (placeholder).