ENH: template kolmogorov for float and double

[fbourgey]

Reference issue

What does this implement/fix?

Replace the separate float and double overloads with a constrained template for kolmogorov. This PR tries to replicate was is done in https://github.com/scipy/xsf/blob/main/include/xsf/tools.h#L45.

Note that https://github.com/scipy/xsf/blob/main/include/xsf/gamma.h#L11 appears to have no constraints.

Additional information

AI Generation Disclosure

Asked Codex for feedback.

[fbourgey]

Another solution is to define in xsf/config.h

template <typename T>
using enable_if_float_or_double_t = typename std::enable_if<is_float_or_double_v<T>, int>::type;

and use

template <typename T, enable_if_float_or_double_t<T> = 0>
    inline T kolmogorov(T x) {
        return static_cast<T>(cephes::kolmogorov(static_cast<double>(x)));
    }

@steppi, @j-bowhay @dschmitz89 any thoughts?