Harvester: avoiding duplicates between systems

[kpsherva]

when harvesting a record from inspire which has legacy recid (integer identifier) the harvester should:
- check if the record is in repository (does it have legacy_recid minted in the pidstore?) if yes - great, proceed with harvesting, if not:

            - check if the record exists in the old system - if it redirects to repository.cern, it means it was migrated already but something went wrong with our pid table, raise an error to check this
            - if it does not exist in the old system - raise an error, means that there is something wrong with the identifiers we get from inspire
            - if it exists but does not redirect - don't harvest the record, add a warning that the record is still in the old system

think of how we could use other identifiers as well to detect and avoid duplication before the record is actually migrated