CALL_THREAD_FCT usage

[Adhemerval Zanella]

While implementing the C11 thread support I noted x86 is the only architecture
that actually defines CALL_THREAD_FCT.  The 32 bits version seems to define it
to explicit align the stack to 16 bytes before call the user provided symbol,
while 64 and x32 seems to just make a function call. Do we still really need
such macro for x86?