Generate a unique id for each occurrence of a tagged term across all three versions of text. Then extract into a 4 column csv (id, tc, tcn, tl) for each term type (e.g. count the number of tags preceding and add one)
see #11