Remove any need for using long-lived secrets from CI

[pivotaljohn]

Describe the problem/challenge you have
We did this:

• Make it easy to run End-to-End tests without hitting into Docker Rate Limiting errors #152
  by plumbing our DockerHub credentials into the workflow.

However, GitHub does not supply the secrets context to workflows running from a fork of a public repo (docs).

So, our workflow failed for such pull requests:

• https://github.com/vmware-tanzu/carvel-kbld/pull/149/checks?check_run_id=3562032285
  on account that the secret is empty and the script initially didn't account for that condition.

This means that contributors that must use a fork of the repo are subject to DockerHub's rate limiting, again.

Describe the solution you'd like
Be able to run a complete end-to-end test without requiring sensitive data.

This could be:

• ensuring that all referenced images are served from somewhere other than DockerHub.
• rewriting all container image references that might point to DockerHub to the registry that's already deployed to the local Kubernetes cluster (via minikube).

Anything else you would like to add:
Bonus points if that registry could be deployed with TLS enabled and secured (i.e. fitted with ephemeral credentials).

There's a test that's currently being skipped on the account of the deployed registry not being secure:
https://github.com/vmware-tanzu/carvel-kbld/blob/7c1bb9b04735ce94d3edfe3faa4f4181e51ebbc3/test/e2e/build_kubectl_buildkit_test.go#L58-L64

---

Vote on this request

This is an invitation to the community to vote on issues, to help us prioritize our backlog. Use the "smiley face" up to the right of this comment to vote.

👍 "I would like to see this addressed as soon as possible"
👎 "There are other more important things to focus on right now"

We are also happy to receive and review Pull Requests if you want to help working on this issue.

[pivotaljohn]

Note https://ttl.sh/ (ref: https://kubernetes.slack.com/archives/CH8KCCKA5/p1630872310132300)

[aaronshurley]

@pivotaljohn Thanks for creating this issue. I'm marking this as important-soon since we're currently skipping a test and therefore have reduced confidence in our releases.