json format deserialization won't work with transformed text

[mssammon]

https://github.com/CogComp/cogcomp-nlp/blob/master/core-utilities/src/main/java/edu/illinois/cs/cogcomp/core/utilities/JsonSerializer.java#L463

This method requires that strings contain only characters corresponding to tokens. For a transformed text, this means a version of that transformation must be maintained and used here, or some alternative method for computing token offsets must be used (e.g., serializing them explicitly)