Manually create a Fargate task with an appropriate compute environment in staging

[lydiascarf]

This is an experiment to replace the batch processing with a Fargate task.

We should:

• create a new Fargate task in the AWS console
  • name the fargate task and associated task something -manual so it won't overlap
• get everything in place to run an analysis using the queue associated with this new task
• confirm that the analysis runs and works
• we may need a thin PR with changes to job definitions, but we don't need to PR if only changes to tfvars