libinput: Use floating point instead of fixed point numbers

Also update configure.ac to require libinput 0.3 when enabled, as it is
the version where double replaced li_fixed_t.

Signed-off-by: Jonas Ådahl <jadahl@gmail.com>
2 files changed