Integrate KiwixStorage

[rgaudin]

Ours scrapers all depend on S3 and use KiwixStorage for it. That wrapper and repo are mostly untouched (yet working) and would great benefit from being integrated into scraperlib: tests, usage, dependencies update, better API, etc.

Now that we had experience using it, it probably make sense to reduce its API to what's actually used.

It's true that some non-scraper projects are using it (cardshop and zimfarm to upload to S3), but those are also using scraperlib and we've discussed making scraperlib's dependencies more modular via extras anyway.