GitLab

This fixes the umac test on win64.

The Lpartial subfunction is entered with plain call instructions,
and the win64 epilogue should only run when actually exiting the
whole salsa20_crypt function.

The stack is not guaranteed to be 16-byte aligned on win64.

This fixes cross building for cases where EXEEXT differs from
EXEEXT_FOR_BUILD, such as when building for windows from unix.

Renamed directory armv7 to arm. New subdirectory arm/neon, for files
using neon instructions. configure.ac hacked to make use of neon
configurable.