Create initial data docker image.

[keyurva]

• This PR creates an initial data docker image that runs the simple importer and the embeddings builder.
  • It does not include other functionality (like versioning) yet.
• The docker images are created here: gcr.io/datcom-ci/datacommons-data
  • The PR includes cloudbuild files to build and push it via a manual trigger (example) and also includes a shell script that should be called from the autopush build (cc @hqpho)
• The Dockerfile does a multi-stage build for better build times and smaller final images.
  • We should adopt this pattern for the services build as well.
• Verified by building and running both locally and in the cloud.
• Here's one I setup as a cloud run job for the dc-dev data
• To run it locally, one can use the following command similar to the services docker (the same env.list file will work for both services and data dockers):

  docker run \
  --env-file env.list \
  # For GCS data.
  -e GOOGLE_APPLICATION_CREDENTIALS=/gcp/creds.json \
  -v $HOME/.config/gcloud/application_default_credentials.json:/gcp/creds.json:ro \
  # For local data.
  -v /path/to/data:/path/to/data \
  gcr.io/datcom-ci/datacommons-data:latest

• @kmoscoe - we can discuss documentation for this when we meet next.