Fix openacc float reduction testcases

Fix openacc float reduction testcases

2017-10-04  Tom de Vries  <tom@codesourcery.com>

	* testsuite/libgomp.oacc-c-c++-common/par-loop-comb-reduction-3.c
	(main): Reduce sum of arr elements.  Assert that hres is exactly
	representable in 32-bit floating point.
	* testsuite/libgomp.oacc-c-c++-common/par-loop-comb-reduction-4.c
	(main): Reduce sum of arr elements.  Assert that hres and hmres are
	exactly representable in 32-bit floating point.
	* testsuite/libgomp.oacc-c-c++-common/reduction-7.c (gwv_np_4): Same.

From-SVN: r253398