Additionally, the pipeline already produces production deployment images for services like blubberoid, mathoid, etc., and is intended to do so for most services in future.
@hashar points out that eventually this will present a storage problem, especially given the number of commits to mw/core (~2118 in last 6 months).
@herron or others, any thoughts on that? Is there any current mechanism or future plans for something that would clean up stale images?
I imagine we'd want to retain production artifacts indefinitely, but dev images should only really need to stay around for some pre-defined time period - 1 week? 1 month? - since people will mostly be using the latest image, already tagged "dev".