GMP 4.4 remaining tasks

[Torbjorn Granlund]

bodrato at mail.dm.unipi.it writes:

  I did not collect timings very carefully, but my first attempt suggest a
  smaller threshold, somewhere around 20 limbs, compared to the mpn_invert
  currently available in the repository.
  
I committed INV_NEWTON_THRESHOLD tuning code, but left it commented out.
Just change line 1915 in tuneup.c to enable it.  (Remember to run
.bootstrap to get the Makefile.am change.)

-- 
Torbjörn