Use fast memcmp() instead of a loop using the slow PyUnicode_READ() macro.
strlen() is still necessary to check Unicode string containing null bytes.