Be able to store and analyze previous failed CI runs to see:
- What part of the CI pipeline failed (eg: provision-ceph, provision-kubernetes, etc)
- How often parts of the CI pipeline failed over a time window
- More statistics to be added as needed?
cc @ben-z
