Get bootstrapping working for computed quantities

[kbenoit]

This will involve resampling from tokenised elements of the original texts, including sentences. Requires pre-tokenisation and indexing of the corpus.