Prefer Docker over Podman in container-image CI scripts

[Copilot]

The Build and publish container image job was building the image with Podman while downstream steps expected to find it through Docker. On GitHub Actions runners with both runtimes installed, that left Trivy scanning a tag that did not exist in Docker’s local image store.

• Root cause

  • ci/build_image.sh preferred Podman when both runtimes were available.
  • The image was committed as localhost/linear-cli-ex:<tag>, while the workflow scanned linear-cli-ex:<tag> via Trivy/Docker.

• Change

  • Reordered runtime selection to prefer Docker first, then fall back to Podman, in:
    • ci/build_image.sh
    • ci/publish.sh
    • ci/save_image.sh

• Effect

  • Build, scan, publish, and save now operate on the same local image representation in CI.
  • The change keeps local fallback behavior for environments that only have Podman.

• Representative change

  if command -v docker >/dev/null 2>&1 && docker info >/dev/null 2>&1
  then
      runtime=docker
  elif command -v podman >/dev/null 2>&1 && podman info >/dev/null 2>&1
  then
      runtime=podman
  else
      die 6 "No usable container runtime found"
  fi

[bougyman]

Your commit messages suck, and we prefer podman, not docker. This was a bad change.