Liveness checks currently kick in after initialDelaySeconds:150, which is much less than the roughly 17m it can currently take to scrape Quay.  If we're too slow to scrape, failing liveness checks can terminate the container just as it was about to be ready, and the replacement container has to start scraping over again.

We should configure a startup probe to allow 20m before the liveness/readiness checks kick in.

We should also reconsider caching scraped information in volumes to recycle between subsequent containers (and possibly subsequent pods): OTA-219.

And also consider improving Cincinnati's ability to detect release images.