Define a strategy for storage of "user data" in readers

https://github.com/openzim/freecodecamp/issues/40 and https://github.com/openzim/freecodecamp/issues/92 require access to "some storage" in the ZIM reader to store data about user progress.

We need to first prototype and define a reference strategy so that what is done inside the ZIM integrates well:

• good support in all (most?) kiwix readers with a stable JS API
• persistent and "unlimited" storage volume:
  • no ephemeral data since it cannot be rebuild
  • no too hard limits on storage volume, even if some MB is probably sufficient in most cases, but could be somewhere in the tens of MBs)
• strategy to make it available across ZIM updates (ZIM ID will change but same data must be retrieved) and potentially across multiple ZIMs (multiple ZIM might share some user data, e.g. preferred language, but also progress on a freecodecamp course which is present in multiple ZIMs)
• strategy to support multiple users on the same reader (we know that users tend to share their device, e.g. in schools)
  • who is responsible to set current user profile ? Is it done at reader level ?
  • how the ZIM code has access to the information, uses this information to store user data

Some of this work has already been done in the scope of Nautilus but we need to make a reference document and implementation out of it, probably with a roadmap (e.g. we can start without the user profiles in a first iteration).