How to avoid rebuilding images between staging and prod?

In Defang monorepo we deploy to staging, and if successful we deploy to prod, but the latter deployment does not rebuild any images. We do this to ensure we actually end up using the images that were tested in staging. If we'd rebuild all services, we might have new surprises and the prod could fail even though staging passed.

When using the Defang CLI, there's not obvious way to do this. Deploying twice, defang up --mode staging followed by defang up --mode production --project-name defang-prod would result in two builds. We could overwrite the staging by not specifying --project-name, but this is also not exactly what we'd want, since it replaces the staging deployment.

• Perhaps we can rely on the image cache to work across different projects? Ie. deploying with --mode production after --mode staging should result in exactly the same image(s).

Start by tracing how the Defang CLI handles defang up, project names, modes, and Docker image building or caching. Compare the staging and production commands described in the issue. Done means production can use the images tested in staging without rebuilding them or replacing the staging deployment.