Add _Float128 spelling for 128-bit float type

According to https://gcc.gnu.org/onlinedocs/gcc/Floating-Types.html,
"On i386, x86_64, PowerPC, and IA-64 other than HP-UX, __float128 is an alias for _Float128"
and we support __float128

Author: smowton

regression/ansi-c/gcc_types_compatible_p1/main.c

@@ -67,6 +67,13 @@ STATIC_ASSERT(__builtin_types_compatible_p(typeof (hot) *, int *));
 STATIC_ASSERT(__builtin_types_compatible_p(typeof (hot), typeof (janette)));
 STATIC_ASSERT(__builtin_types_compatible_p(__int128, signed __int128));
 
+#ifndef __clang__
+// clang doesn't have these
+#if defined(__x86_64__) || defined(__i386__)
+STATIC_ASSERT(__builtin_types_compatible_p(__float128, _Float128));
+#endif
+#endif
+
 /* Incompatible types */
 
 STATIC_ASSERT(!__builtin_types_compatible_p(char, _Bool));

src/ansi-c/scanner.l

@@ -466,7 +466,8 @@ void ansi_c_scanner_init()
                     return make_identifier();
                 }
 
-"__float128"    { // clang doesn't have it
+"__float128" |
+"_Float128"     { // clang doesn't have it
                   if(PARSER.mode==configt::ansi_ct::flavourt::GCC)
                   { loc(); return TOK_GCC_FLOAT128; }
                   else