The only actual semantic difference is how possible IATA codes in the text
are interpreted for disambiguation. With that now being done only if the
found code is also found as a candidate from the rest of the text we don't
actually need the separation at all anymore.