Compiler warnings

[johnmcfarlane]

Purely a suggestion but if you can get your implementation to compile with warning levels elevated, that would be helpful. I use -Wall -Wextra -Wpedantic -Werror.

/home/john/ws/math/cnl/include/cnl/_impl/fixed_point/../../_impl/math_20.h: In instantiation of ‘constexpr const double cnl::math::e_v’:
/home/john/ws/math/cnl/include/cnl/_impl/fixed_point/../../_impl/math_20.h:75:44: required from here
/home/john/ws/math/cnl/include/cnl/_impl/fixed_point/../../_impl/math_20.h:61:71: error: conversion from ‘long double’ to ‘double’ changes value from ‘2.71828182845904523543e+0l’ to ‘2.7182818284590451e+0’ [-Werror=float-conversion]
61 | template inline constexpr T e_v = 2.718281828459045235360287471352662498L;
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[johnmcfarlane]

A fix could be:

      template <FloatingPoint T> inline constexpr T e_v<T>          = T(2.718281828459045235360287471352662498L);

Let me know if you'd like a PR.