Document building the Cloud ML container

[jacobtomlinson]

We are stopping building and publishing the Cloud ML container. Perhaps we should document how to build it instead for the small number of folks who do use it.

[jacobtomlinson]

Now that we are making good progress with #203 and are switching out the Cloud ML container for the base RAPIDS container and installing a few extra packages at runtime it's not actually becoming much of a burden.

Most notebooks only need one or two extra packages to be installed and we can either do it inline in the notebook or by using the EXTRA_PIP_PACKAGES and EXTRA_CONDA_PACKAGES env vars when spinning up other resources.

So I expect as we finish up this notebook migration the usage of the Cloud ML container will drop to zero. So I don't think we should bother documenting building this container.