Add missing type traits about core

[bruxisma]

common_type was implemented as it would cause an issue if given a typelist of void (even if every type was void) under libc++. Because common_type is used by the variant<Ts...> for visitation, I needed one that worked for both libc++ and libstdc++.

I should note that a majority of them were not implemented for reasons where the ::type typedef would be the actual type itself. For instance, typename std::is_constructible<T, Args...>::type is the same as std::is_constructible<T, Args...>. Additionally because I can't do something like template <class Args...> constexpr auto is_constructible_v = std::is_constructible<Args...>::value, I avoided that as well.

While some types do have the _t alias available (such as: make_signed), many of the others (such as is_null_pointer, which was added after work began on this library) do not have an equivalent alias.

I apologize if this is an inconvenience, but I will not be adding a type alias for types that resolve to std::true_type or std::false_type.

However, I'll not close this yet as is_null_pointer definitely needs to be added, and will be for 1.1.

(Please note that is_final will not be added as that is a compiler specific trait and having that specific trait suddenly break because a given compiler wasn't supported is not something that a user would enjoy)

[tarqd]

Ah I see your point. I was more so suggesting it for completeness. Having the _t alias might also reduce confusion (you can just use the aliases for everything, versus taking a few brain cycles to think of which one to use).

As for common_type, is that related to this?

[bruxisma]

The common_type situation I believe was something more where T would be turned into T&&, and since you can't declare a void&&, this would cause clang on OS X to perform its best Wu-Tang Clan impression, and spit words endlessly. I do not believe that MNMLSTC Core's common_type is truly usable with SFINAE (try passing in a few lambdas to variant<Ts...>::match and watch the fireworks), but it was certainly not related to that proposal.

[bruxisma]

is_null_pointer has been added. I will now close this issue.