IssueTracker

As written in [1], I'm creating a separate issue to ask for an improvement of GAR cleanup policies. Specifically, I'd like to specify cleanup policies based on when an artifact was last accessed / downloaded. Existing policies in fact don't prevent a possible production outage:
- A deletion policy on olderThan may delete an artifact (e.g. a Docker image) that is used in production
- This could be partially prevented by combining with a "mostRecentVersions" policy under the assumption that the production artifacts are among the last N uploaded to GAR. However, if this assumption fails, a production outage will happen anyways.

A robust way to solve this is having a policy that deletes an artifact only if it hasn't been accessed in the last N days

[1] https://issuetracker.google.com/issues/215403088
[2] https://cloud.google.com/artifact-registry/docs/repositories/cleanup-policy