GMP 5.1.1 make check fail on Solaris 11

[Torbjorn Granlund]

Marc Glisse <marc.glisse at inria.fr> writes:

  I think they are only used in this test program to check that
  double->mpf->round gives the same result as the libm rounding
  functions.

OK.

Perhaps it would be better to make some consistency checks, like ceil(x)
>= x >= floor(x), ceil(x)-1 < x, floow(x)+1 > x, etc.  Not to avoid libm
deps, but to allow testing of a greater range.

-- 
Torbjörn