Add option to `OptimadeClient` to automatically cache results to an OPTIMADE JSONL file

[ml-evs]

Ideally a multi-provider query would write these results in one file per database, storing the query for future use, and the client could look up the final entry in the JSONL in order to restart the query.

[ndaelman-hu]

So, if I understood you correctly, any query would write to a local .temp folder, with a sub-folder structure "query -> provider -> database". This local cache would then be searched through first when the same / a similar query is submitted.

I guess I could start with support for a verbatim query, and we see how to escalate similarity from there? I guess additional filters can be readily applied on top of the cached queries.