Cache CI container dependency layers in GHCR.

[gerrod3]

Split the image build so third-party deps can be reused across runs, and have nightly push the cache for PR jobs to pull.

https://redhat.atlassian.net/browse/PULP-2237

📜 Checklist

• Commits are cleanly separated with meaningful messages (simple features and bug fixes should be squashed to one commit)
• A changelog entry or entries has been added for any significant changes
• Follows the Pulp policy on AI Usage
• (For new features) - User documentation and test coverage has been added

See: Pull Request Walkthrough

[gerrod3]

This approach did not work. The uv pip install is already a magnitude faster than the FROM base-image step that caching the dependencies gives no real speedup, just a lot more complexity in the CI.

Future things to look into:

• S3 runner's custom botocore dependency is a one line change, that can probably just be a sed call so that uv doesn't have to slow down to fetch the github repo.
• S3 also installs ansible.aws which is slow. Look into just creating the minio bucket with shell commands.
• Speed up the one-in-all container startup scripts.
• Reduce the sleep duration in between startup polling.