Use C11 *_DECIMAL_DIG macros in libm-test.inc.

Use C11 *_DECIMAL_DIG macros in libm-test.inc.

Now that we build with -std=gnu11, we can use C11 <float.h> macros
such as FLT_DECIMAL_DIG instead of the GCC predefines such as
__FLT_DECIMAL_DIG__ that are used internally in <float.h>.  This patch
makes libm-test.inc do so.

Tested for x86_64 and x86.

	* math/libm-test.inc (TYPE_DECIMAL_DIG): Use LDBL_DECIMAL_DIG,
	DBL_DECIMAL_DIG and FLT_DECIMAL_DIG instead of __DECIMAL_DIG__,
	__DBL_DECIMAL_DIG__ and __FLT_DECIMAL_DIG__.