Add data corrections argument

[organisciak]

There would be value in an optional cleaning flag that applies opinionated but uncontroversial OCR corrections to the tokens. e.g.

• Stripping trailing non-word characters in ^\W\w+ or \w+\W$ -> Words that have punctuation attached to the first or last index, like "token or token—
• normalizing the medial s unicode (ſ) to a regular s
• separating out ligature characters (e.g. fl ff ffi ffl, Ꝏ ꝏ)