* Implement text filters
** implement abstractize number filter which will take any number as imput and return a NUMBER abstract.
** Implement the abstractize roman numeral filter which will take a token that looks like a roman numeral (except for I) 
   return a NUMBER abstract.
** implement punctuation stripper / abstractizer which will take a punctuation token and return an abstract or abort the
   token.
** Implement asciifier token which will take a unicode string and return an ascii approximation.
* Implement some example code.
* Complete documentation and introductions to actually be useful to users.