Add two compile time options: HAVE_UNICODE_WCHAR_CACHE and
USE_UNICODE_WCHAR_CACHE.
USE_UNICODE_WCHAR_CACHE makes the interpreter not using the
wchar_t cache and the legacy Unicode C API.
HAVE_UNICODE_WCHAR_CACHE removes the wchar_t cache and the
legacy Unicode C API that depends on it.
Do not use the legacy Unicode C API which depends on the wchar_t* cache if USE_UNICODE_WCHAR_CACHE is defined as false.
It can help to test in advance the effect of removing the wchar_t* cache.
https://bugs.python.org/issue36346