feat: Support building and pushing container images shared within a dataset folder

[adlersantos]

Description

Based out of a discussion on #5.

We need to support operators such as KubernetesPodOperator and other GKE operators that allows tasks to be run in specific clusters to isolate resource usage from other data pipelines.

As a prerequisite for such tasks, Docker images need to be defined, built, and pushed to GCR. This PR adds a workflow for building and pushing images.

1. Create an _images folder under your dataset folder if it doesn't exist.

2. Inside the _images folder, create another folder and name it after what the image is expected to do, e.g. process_shapefiles, read_cdf_metadata.

3. In that subfolder, create a Dockerfile and any scripts you need to process the data. Use the COPY command in your Dockerfile to include your scripts in the image.

The resulting file tree for a dataset that uses two container images may look like

datasets
└── sample_patterns
    ├── _images
    │   ├── get_google_cloud_urls
    │   │   ├── Dockerfile
    │   │   └── script.py
    │   └── ping_google
    │       ├── Dockerfile
    │       └── script.py
    ├── _terraform/**/*
    ├── containerized_pipeline/**/*
    └── dataset.yaml

Now, running scripts/generate_dag.py will build and push the containers automatically.

Here's a screenshot of a successful run on the Airflow webserver (Cloud Composer):

Checklist

• Tests pass.
• Linters pass.
• Please merge this PR for me once it is approved.
• If this PR adds/edits/deletes a feature, I have updated the README accordingly.

[leahecole]

Some questions

• How will we be expecting users to represent the requirements for the scripts that will be in their images? (requirements file? Pipfile?) Are we going to be prescriptive, or is that out of scope? Either way, we may want to mention it and show an example of our preferred way, because their Dockerfile will need to copy in those requirements and do the installation path as welll