Supports for hyphen as alternative spans

[Joxit]

Background

Sometime, parsing fails when words are not well split. Hyphens' main purpose is to glue words together. That meas, when an hyphen is used, we can process it like a simple space in order to have two separate words.

Only processing hyphens like spaces can unfortunately not be the final solution because the hyphen is also useful in some other cases.

That's why I suggest to take advantage of our graphs and add some alternative ways to complete a phrase without hyphens.

How it works ?

When we split all sections, we do a first compute on spaces only (like before) and then a second compute on hyphen.

Example for 10 Boulevard Saint-Germain Paris, when we split this section, we get this: 10, Boulevard, Saint-Germain, Paris, here is the graph:

With the hyphen step, we will have 10, Boulevard, Saint-Germain, Paris, Saint, Germain

Thanks to this, we will be able to parse phrases such as :

• 10 Boulevard Saint-Germain Paris: which is housenumber + street (first solution without this PR 👎)
• 10 Boulevard Saint-Germains Paris: which is housenumber + street + locality (first solution with this PR 👍)

[missinglink]

Wow very cool.

I'm trying to wrap my head around all the code since it's quite complex, but I think it looks good.
I've added a couple of minor comments but looks good 👍