Elastic scheduling

[pearce8]

Currently, in a config directory, would template the yaml to submit to EKS. ramble on while running on laptop, could use a templated container for kubernetes. Then the EKS job installs Spack and Ramble into the kubernetes container. There are commands to install spack and Ramble that inject commands.

There are also ramble deployments (push and pull) so Ramble can take an activated workspace and group them into an artifact repository to push to an s3 bucket; then the EKS job would make a workspace and pull from the same job that you pushed.

Later, there will be a feature to have workflow managers as well as package managers, to make it a one-line change to run on EKS instead of slurm.

There may be overlap between functionality we need for containers/eks/sandboxed machines.