Filter to remove failed jobs a few days (configurable) after a newer instance of same job succeeds

What would be an appropriate pattern to implement a job filter such that it sets the expiration wen/once a subsequent instance of the same job succeeds.... i.e. auto-expire failed jobs once the job runs successfully again.

No file, test, or entry point is named. First identify Hangfire's job-filter and expiration extension points, then determine how to recognize a later successful instance and expose the expiration delay as configuration. Done means failed jobs are automatically expired after a configurable delay when a newer instance of the same job succeeds.

Written by the indexing model from the issue text.