This new version allows to compute odd and even double factorials
without harming performance for the standard factorial.

Most operators do not yet handle reals of different sizes, but
this change fixes the most glaring bugs.

It was hard to avoid some of the compiler optimisations caused by our
intermediate conversion to double, so we hide them behind the actual
real object. Also, this commit fixes a potential exponent overflow.

Now that the exponent is no longer biased, it is a signed number, and we
need to account for that when dividing it and expecting rounding rules to
behave in a certain way.

Some checks are failing, I probably messed up several functions.

This is the beginning of the removal of lol::String which is not really
useful once all its tiny utility functions are implemented for std::string.

This allows us to manipulate polynomials of polynomials. Not sure why yet,
but it could be useful to implement 2-variable polynomials.

This gives us a reasonably cross-platform way to to build options. Works
well with LolRemez, where we no longer depend on GUI libraries.

Rename config-build.xml to build.config, move it to the root directory,
and add MSBuild flags for SDL, FFmpeg (unused yet) and OpenGL.

This flag disables GL, GLES, but also Glew and EGL libraries.

When built as a subproject, we need the master to decide what features
to activate or not. This could lead to reduced dependences.

This patch also allows to build without imlib2.

Linux-only for now. This will be necessary for projects that do not
actually use the rendering engine, such as LolRemez.

This will let us detect problems with unused return values. For instance,
if you don’t remember that normalize(q) returns a pointer and expect it
to modify the object, you’ll get this warning:

warning: ignoring return value of function declared with 'nodiscard' attribute [-Wunused-result]
    quat q(1,2,3,4);  normalize(q);
                      ^~~~~~~~~ ~

From now on, msg::debug will not display anything, unless this is a debug
build, or the LOL_DEBUG environment variable is set.

By adding a bias we ensure that 1.999… will be displayed as 2.0 as long as we
do not ask for too many digits, and that rounding is reasonable for most values.
Note that our need for proper rounding is not as high as for IEEE floats since
we have an insane amount of digits at our disposal.

Also we now get rid of trailing zeroes when printing reals.