Add document hashes.

[J08nY]

Fixes #220.

[adamjanovsky]

Not sure if it's wise to add also txt document hashes, dunno how deterministic the conversion tools are. But we can try and we'll see I guess.

[J08nY]

So with the txt hashes that is exactly what I wanted to capture, we already have several heuristics in the tool that could change without the underlying data changing and the pdftotext is one of them, so we at least know when we change stuff.