Run image builds on in-cluster k8s executor

What does this MR do?

The VPN connection is flaky, and I regularly have builds timing out or failing due to it. By running the image build jobs in our cluster, we can skip the VPN and connect directly to Orka.

I also had an issue with High Sierra where the node asdf plugin fails to verify signatures. I can't figure it out and this OS is really not a priority so I'm just working around it for now. Also had an issue with host key checking that wasn't properly disabled and triggered when we ran ansible multiple times on the same host (rather than losing known_hosts between runs on the docker runners). Finally, I reverted the linters to their state pre-dogfooding before !57 (merged)

Also did a bunch of additional cleanup, with unused variables, closed beta provisioning script update, etc.

What's the best way to test this MR?

Builds should be passing, no change should be made to the image contents.

What are the relevant issue numbers?

Closes #63 (closed)