Various improvements to AWS Batch support

[gillins]

This PR adds the following:

• Allow setting of the AWS Region used for creation of resources
• Insert the region in the bucket name so a user can have buckets in different regions without name clash
• A new ECR repo for the "main" script that actually contains the user's script. Building and pushing to this (along with the submission to Batch) is left as a task for the user. It seems more likely that users will want to run their "main" script as a Batch job rather than running locally.
• Permissions to allow submission of jobs so that the user's "main" script as a Batch job can then fire off worker jobs.
• Implement tags for users

In future I suspect it may be necessary to ensure that the "main" script has access to more local storage as you'd only be using this if you had very large files and by default Batch jobs don't have much local storage. This can be done by either adding a LaunchTemplate to the ComputeEnvironment that increases the size of the root volume, or my directly creating and mounting an EBS Volume. RIOS should probably do something here to make it easy, but I'll make that a separate PR.

[neilflood]

That seems all very good (as if I understood it....). My only quibble is to request that you change the name of the region environment variable. Currently you have it as RIOS_BATCH_REGION, but I would prefer RIOS_AWSBATCH_REGION, just to be completely clear that this relates only to AWS Batch.

Sometimes my nit-picking amazes even me........

[gillins]

Good thought!

[neilflood]

I just bothered to check, and it appears you already had a RIOS_BATCH_STACK variable as well. Sorry, I had not noticed that. Anyway they should be consistent, so I guess that means change that to RIOS_AWSBATCH_STACK. Does it???

Sorry, I am not being a very good reviewer.

[gillins]

I'd forgotten about that - good spotting. PR's are great aren't they.