Feature/optimize circleci pipeline

[nwang92]

Switching up the CircleCI pipeline a bit. Summary of changes:

• There's some lengthy steps where we save images as artifacts - to my knowledge, we're not using them so it's a bit of a waste
• Added clair-scanner into CircleCI instead of the Makefile, so we (potentially) don't need our own clair-to-junit parser
• Ordering pipeline so we build images once and share with downstream steps (vs building on each step)
• Support for different instance types to get large tests to run faster
• Fixing/adding tests

Anyways, the previous pipeline would take 1.5h, and now it just takes ~1h.

[bb03]

Is there any parallelism here? Should we run these distros in parallel to try and speed up the CI?

[nwang92]

Yeah, they already are: https://app.circleci.com/pipelines/github/splunk/docker-splunk/938/workflows/aaf37ec1-d0b3-49e9-8f65-eac16cb260fa