Splitting languages into loadable dictionaries and adding Swedish language

[adrael]

Hello @sindresorhus !

This PR improves the management of languages through dictionaries.
Users can now explicitly load the dictionary-ies they need to properly slugify their data.

This PR was needed because of languages overlapping.
For example, Swedish and German have common chars (such as Ä), and thus the wrong replacements were made.

It is now fixed, and thus fixes #33

Also, this PR adds support for a lot of new languages, with according tests, to amend long-pending PRs:

• armenian
• czech
• dhivehi
• georgian
• greek
• hungarian
• latin
• latvian
• lithuanian
• macedonian
• polish
• serbian
• slovak
• ukrainian

Thanks for the good work!

[sindresorhus]

Instead of the user having to manually load the wanted dictionaries, I think it would be better if the user could just provide a language hint, for example, sv-SE to prioritize the swedish replacement. And adding Swedish should be a separate PR from splitting up the languages. And sorry for the late reply. I just got busy.