C++ constexpr experiments

[Shelwien]

test-crc.cpp: compile-time crc32 calculation
https://godbolt.org/z/vCT5tA


test-format-str.cpp: compile-time printf format string generation based on arg types
https://godbolt.org/z/cCmYTi


test-concat.cpp: compile-time string concatenation
https://godbolt.org/z/gF5_y7

[Shelwien]

test-concat2.cpp: compile-time string concatenation attempt #2
https://godbolt.org/z/IszYEH

Previous attempt actually created _an initializer_ for a string.
Sure, gcc optimizes it to copying of an actual static string...
but IntelC doesn't - it generates a sequence of byte writes which construct the string in runtime.

Well, now there's an actual static string which isn't copied anywhere.

The main problem: its apparently prohibited to use constexpr on function arguments,
so any constexpr var passed to a function turns into not-constexpr.
Ok, but its possible to create a local class with constexpr methods
which return these arguments, and pass its instance to a function.

Problem #2: to let function call still look normal, we can make a macro wrapper
which would automatically convert arguments into return values of constexpr methods.
Gcc has a problem with creating an anonymous instance of anonymous struct inplace,
like func( struct{}() ), but somehow its not a problem if its an inplace lambda returning that struct.

Problem #3: this specific source works in gcc,intelC,clang, but not in VS -
looks like VS can't access symbols of a constexpr string as constexpr.
Well, its not like anybody uses it... :)