Buildfix: include <cfloat> for FLT_MAX in TestArm64Emitter

MSVC pulls it in transitively, gcc and clang don't - broke the gcc-normal,
clang-normal, macos and test-headless-alpine CI jobs.

unittest/TestArm64Emitter.cpp

@@ -4,6 +4,7 @@
 #include "Core/MIPS/MIPSVFPUUtils.h"
 #include "Core/Util/DisArm64.h"
 
+#include <cfloat>
 #include <string.h>
 
 #include "UnitTest.h"