There are more edges now: when the xml refers to other parts of the xml by means of ids, we turn that into edges.

And we added phrases and clauses.

• Attributes that only ever have the value true get value 1 (int).
• Generated numbers for books, sentences, words are now all in feature num.
• Features are not split between the w and wg elements, if they have values for both, they have slightly different meanings
• The descriptions of the features are a little bit clearer

This version of the data has the words in proper order.
Achieved by a new version of the TF walker converter, which can now reorder slots.

Data version 0.1.1