Implement a schema field of type tokens

[nikhilk]

This is for representing tokenized set of values (eg. comma separated strings or numbers). The ideas is as part of data preparation, one can convert text into a set of tokens (after doing word breaking, stop-word removal, stemming etc.)