api: support finer-grain datetime

[kaplun]

Currently, according to https://claimstore.readthedocs.org/en/latest/users.html#list-claims, the claimstore allows for datetime-based queries of the type:

• since (datetime) – it must have the format ‘YYYY-MM-DD’. It fetches claims that were created from this given datetime.
• until (datetime) – it must have the format ‘YYYY-MM-DD’. It fetches claims that were created up to this given datetime.

The day-based granularity could not be enough for certain services which needs to exchange and query IDs more often during the day.

Already today, we are exchanging between CDS and INSPIRE, respectively CDS IDs and INSPIRE IDs for corresponding papers, and due to limitation in OAI-PMH granularity and other technical constraints this propagation happens with potential 4-days of delays. With some hacks we are managing to put it under one day.

Would be great if no hacks were needed to have less-than-a-day updates once we move to exchange IDs via the claimstore.

[tiborsimko]

Hmm, each service can use REST API of the central ClaimStore to push their claims (or interrogate others' claims) many times during the day at their leisure according to their "live" needs. Hence I wonder whether it isn't enough to query about some wanted PID when a need arise and relying on getting a "real-time" replies back? The REST API paradigm provides here quite a difference already when compared to exchanging OAI-PMH feeds between two parties.

(That said, if there is a need to restrict resulting claims in time to those issued between, say, 10:00 and 11:23, then we can surely add that...)

[kaplun]

I can imagine it is more efficient for a client to query the claimstore in bulk for all the new claims related to some namespace since the last time (with general granularity) the query was performed, rather than perform thousands of get reqeusts for every known ID.

E.g.: Give me anything new has been declared since yesterday 10:33:11 on INSPIRE records.