Generate and include test data

[bencomp]

I have been building and testing the code using a very small set of converted articles. Because these articles are protected by copyright, I cannot include them in the repository.

It would be great if the test data could be shared. To make sure that shareable test data are representative:

• use current test files, but replace words in the body and title with random data
• keep the markup of search phrases
• keep the capitalisation
• keep length of words the same
• keep punctuation
• words replacements may be the same within a document, but
• randomisation may not be reversible

Perhaps this is something for https://github.com/gchq/CyberChef?