Issue #17615: On Windows (VS2010), Performances of wmemcmp() to compare Unicode

strings are not convincing. For UCS2 (16-bit wchar_t type), use a dummy loop
instead of wmemcmp(). The dummy loop is as fast, or a little bit faster.

wchar_t is only 16-bit long on Windows. wmemcmp() is still used for 32-bit
wchar_t.

Objects/unicodeobject.c

@@ -10375,16 +10375,7 @@ unicode_compare(PyObject *str1, PyObject *str2)
             break;
         case PyUnicode_2BYTE_KIND:
         {
-#if defined(HAVE_WMEMCMP) && SIZEOF_WCHAR_T == 2
-            int cmp = wmemcmp((wchar_t *)data1, (wchar_t *)data2, len);
-            /* normalize result of wmemcmp() into the range [-1; 1] */
-            if (cmp < 0)
-                return -1;
-            if (cmp > 0)
-                return 1;
-#else
             COMPARE(Py_UCS2, Py_UCS2);
-#endif
             break;
         }
         case PyUnicode_4BYTE_KIND:

PC/pyconfig.h

@@ -645,9 +645,6 @@ Py_NO_ENABLE_SHARED to find out.  Also support MS_NO_COREDLL for b/w compat */
 #define HAVE_WCSXFRM 1
 #endif
 
-/* Define to 1 if you have the `wmemcmp' function. */
-#define HAVE_WMEMCMP 1
-
 /* Define if the zlib library has inflateCopy */
 #define HAVE_ZLIB_COPY 1