ESGF search records contain various ids. These include:
- id: internal identifier so that SOLr can keep track of each record. Currently the instance_id + index node
- drs_id: DRS identifier without the version number. Effectively an identifier for each dataset (including all versions and replicas)
- master_id: The same as drs_id. I think this was created because of confusion about the purpose of DRS and drs_id. E.g. it's not clear whether a DRS identifier includes a version
- instance_id: <master_id>.. I.e. the DRS including the version number
pyclient adds to this confusion be exposing the id as Result.dataset_id. This needs to be changed in some backward-compatible way. We need a sane naming scheme for identifiers.