GrouperLoader: Cleanup based on Loader Metadata

We have begun creating auxiliary/sidecar groups based on memberships of other groups, sometimes reference groups. For example, we have Admin (aka PSA) accounts issued to some people in addition to their personal account, and we sometimes want to group those Admin accounts. Another example is where we want the Supervisors of people in a Group (eg, Supervisors of people who have not completed mandatory training so we can notify those supervisors).

Again, these use cases and others have led us to selectively create these derivative groups, and the best location we've determined for them is adjacent to the source group.

The (only?) downside of this approach is that the we end up have multiple loader jobs creating groups in a given folder... ref:affiliations may be 99% managed by our affiliation-loading job, but a handful of groups are added and managed by these sidecar capabilities. This means that the affiliation-loading job can not 'own' ref:affiliations to clean up old groups when a department goes away or whatever.

This Jira is requesting that sql-group-list loader jobs (or other multiple-group-managing loader jobs) support removing groups they previously loaded (probably based on loader metadata) when they're no longer listed in the upstream system. We think this should be the default behavior, but understand that that might be dangerous at this point.