Read / write int 8 / 16 / 32 / 64 / 128 functions

In low-level code one sometimes has to read or write integers that may be unaligned, may be big endian, little endian or native. Could functions for this be provided? The non-native ones are non-trivial.

IMO the language itself should provide these as intrinsics but until then..

A reinterpret_cast isn't sufficient: https://github.com/abseil/abseil-cpp/issues/269