Add functionality to clean duplicate Practitioners from the HAPI server

[Wambere]

IMPORTANT: Where possible all PRs must be linked to a Github issue

Fixes https://github.com/onaio/fhir-tooling/issues/58

• Checks if keycloak user has Practitioner(s) linked to it
• If only 1 is linked, it confirms that it is the same uuid provided in the csv
• If more than 1 is linked, it attempts to delete all Practitioners with uuids not on the csv
• Optional cascade delete available

[ageryck]

it turns out cascade_delete only cascades for Patient resources, we may need to find a way to get linked resources id's and delete them as batch via the transactional delete

[Wambere]

this seems very scary if the practitioner has linked data, whether cascade_delete is true or false.

when would we expect this to be used? Once data collection has started we probably want to clean up links instead of delete data.

@pld I agree, technically this should not have happened in the first place. I think these "extra" resources were created in the very beginning before we started passing in uuids for Practitioners to the csv. So when we ran the script again with new uuids, we created duplicates. Also I think passing cascade_delete as false (at least for the first run) is safe-ish since anything with related resources will not be deleted, and anything that is actually deleted is logged and is only soft deleted and can be recovered

[pld]

this seems very scary if the practitioner has linked data, whether cascade_delete is true or false. when would we expect this to be used? Once data collection has started we probably want to clean up links instead of delete data.

@pld I agree, technically this should not have happened in the first place. I think these "extra" resources were created in the very beginning before we started passing in uuids for Practitioners to the csv. So when we ran the script again with new uuids, we created duplicates. Also I think passing cascade_delete as false (at least for the first run) is safe-ish since anything with related resources will not be deleted, and anything that is actually deleted is logged and is only soft deleted and can be recovered

OK, can you please add these details on when to use it to the readme? And a confirmation prompt when it's called from the cli

[ageryck]

it turns out cascade_delete only cascades for Patient resources, we may need to find a way to get linked resources id's and delete them as batch via the transactional delete

This worked for @Wambere , it could be a permission issue on the target server