Consider unifying _ntoa_long and _ntoa_long_long

[eyalroz]

When ll is supported, why should we have both _ntoa_long and _ntoa_long_long? They do the exact same thing, except for the different types. Why not just put the value in a long long and use just one of the functions? It's not as though we care that much about the performance cost or stack size along the long path that saving 2 or 4 bytes matters so much... after all, we don't have _ntoa_short.

@mickjc750 : This is your idea originally.

[eyalroz]

Well, it's done on my fork, that wasn't so hard...