Language tweaks like converting "who's" to "who"

[marks]

Just one of the small caveats I see when applying this downstream from speech recognition. I plan on diving deeper into the @CivOmega codebase later but if someone could point me in the right direction of where to implement this, that'd be awesome)