Containerize applications

As mentioned in #2 , production applications should be deployed using containers to make them more portable and accessible to users without dedicated resources. As I discuss in fastmachinelearning/gw-iaas#17, building a simplified framework for container builds in the context of the monorepo format with Poetry is not exactly trivial, but for now at a minimum applications should include a Dockerfile and an associated GitHub workflow which builds the container, runs tests inside of it, and pushes it to an ML4GW repository.

• Where do we want this repository to live? I assume on DockerHub? Should we create an organization on there as well?
• If yes, we'll need to include DOCKERHUB_USERNAME and DOCKERHUB_API_KEY secret variables on our build workflows using a GitHub environment
• We'll need to follow these instructions to make our containers available on LDG
• We'll want to include two tags for each container: latest and stable (or something similar). latest will be pushed from the most recent build of the code, while the build workflow should pull the last latest and then push it to stable as a backup in case something breaks with latest. The DockerHub repo should maintain timestamped tags (e.g. monthly 22.01) for all versions