more classification/traits

For more precise classifications it is still useful to provide these traits:

* is_character -> char32/char16/char8/wchar/char
* is_integer -> signed and unsigned int128..8
* is_unsigned_integer -> is_integer && is_unsigned
* is_signed_integer -> is_integer && is_signed
* is_bool -> only bool
* is_logic -> is_bool || boost::tribool (specialized for tribool in boost::logic)
I intend to provide a 4-valued logic (https://en.wikipedia.org/wiki/Many-valued_logic#Belnap_logic_(B4)) and a more modern variant of a 3-valued logic (instead of boost::tribool) (and for completeness "normal 2-valued logic"):
* is_logic -> is_bool4 || is_bool3 || is_bool2 || is_bool
* is_numeric -> is_floating_point || is_integer
* is_function_pointer (like is_member_function_pointer) -> is_pointer && is_function
* dependent_false -> "always false" idiom: template struct dependent_false : public false_type{};
* is_quotrem_order -> is std::Xdiv_t implemented as {quot, rem} or {rem, quot} ? https://en.cppreference.com/w/cpp/numeric/math/div
This is needed at https://en.cppreference.com/w/cpp/language/structured_binding
auto [a, b] = std::div(x, y)
-> is a quot or rem or is b quot or rem ?
(PS: maybe you could submit a proposal to the C++ standard committee to set Xdiv_t to a unique order).

exact/specific traits eg
* is_char32, is_wchar, ...
* is_uint64, is_int16, ...
* is_std_longdouble/double/float
* is_std_float128/64/32/16, is_std_bfloat16
* is_boost_float128/80/64/32/16
* is_float64 -> is_std_float64 || is_boost_float64, ...

thx
Gero