glibc: cleanup long double support

- Add 128 bit long double support from current glibc and a few headers
they need
- Move the existing 80 bit float support in a sub directory of generic,
it is

glibc: cleanup long double support

- Add 128 bit long double support from current glibc and a few headers
they need
- Move the existing 80 bit float support in a sub directory of generic,
it is not universal to all archs (see file added in docs/develop/arch).
Also include some new .h files for x86 that are needed after these
changes (from newer versions of the glibc).
- Adjust Jamfiles for m68k, x86 and x86_64 to use the 80bit format
- Do not adjust arm jamfiles, it was wrongly using 80bit long double and
should be fixed to use 64bit instead (which means the double functions
can be used with aliases)
- Do not adjust powerpc jamfiles, because it uses yet another format and
we build it without long double support anyways.

Note that I moved only the files that were creating compile errors,
quite likely more of the s_* and e_* files need to be moved to the
specific directories, see glibc list here:

https://sourceware.org/git/?p=glibc.git;a=tree;f=sysdeps/ieee754/ldbl-128
https://sourceware.org/git/?p=glibc.git;a=tree;f=sysdeps/ieee754/ldbl-96

Change-Id: Ic2d8a454ba9a3b99638e4fbb63daf02df0fea403
Reviewed-on: https://review.haiku-os.org/c/1143
Reviewed-by: waddlesplash <waddlesplash@gmail.com>

show more ...