Introduce a dedicated scalar kind for characters
Characters should be interpreted neither as signed, nor unsigned integers, but rather according to their encoding. We should change the way this is currently handled in PDI.
Another different & more complex issue regards the support of variable-length types such as UTF8 characters...