Build the cache system

[kelson42]

We don't want to create unnecessarily large amount of requests on upstream servers. Therefore we probably need a cache for the scraper.

Our usual cache storage is ObjectStorage.

Before starting with implementation we should decide:

• Does a cache makes sense?
• What should be cache?
• How works cache invalidation?

[benoit74]

We won't need a cache for v0.1

We might want to optimize perf at Zimfarm level by keeping downloaded files over runs, but this is only an optimization, not needed per se and not a scraper feature, already feasible for people running the scraper locally.