support FLT_EVAL_METHOD changing on x86 with gcc -msse2 -mfpmath=sse

if the compiler provides a value, use it; otherwise fallback to the
platform default (2).

arch/i386/bits/float.h

 #define FLT_ROUNDS 1
+#ifdef __FLT_EVAL_METHOD__
+#define FLT_EVAL_METHOD __FLT_EVAL_METHOD__
+#else
 #define FLT_EVAL_METHOD 2
+#endif
 
 #define LDBL_MIN     3.3621031431120935063e-4932L
 #define LDBL_MAX     1.1897314953572317650e+4932L