BLD: Split Docker builds into separate jobs

[qinxuye]

Summary

• Split Docker CD image publishing into separate aarch64, GPU, and CPU build jobs while keeping the original sequential order.
• Add per-job Docker cache cleanup before and after each build.
• Move latest tag publishing into a separate job that runs after all image builds succeed.
• Open or update a GitHub issue automatically when Docker CD image publishing fails.

Tests

• python yaml.safe_load for .github/workflows/docker-cd.yaml
• git diff --check

[gemini-code-assist]

Note

Gemini is unable to generate a review for this pull request due to the file types involved not being currently supported.

[rogercloud]

Reviewed the Docker CD workflow split at head 9893835. I did not find any blocking issues in the workflow changes. The currently failing metal/GPU CI jobs appear unrelated to this Docker CD-only change.