Improve verify-range testing

[2opremio]

While testing captive core with verify-range in AWS batch I encountered a few problems which made the experience unpleasant:

1. Manually editing the Job Definitions/Queues/Compute environments in the AWS console was hairy and error-prone.
2. Spawning jobs suffers from (1) aggravated by the need to calculate the ledger ranges (we have a spreadsheet for that, but still)
3. When a job fails, there is no way to simply respawn jobs. You need to manually create a new job by copy-pasting the fields of the failed job (increasing the number of retries is often not enough).

To improve the situation, I propose the following:

1. Use Terraform to keep track of the definiton AWS Batch resources and apply them automatically on merge. This will probably require having a separate repo for verify-range.
2. Spawn test jobs semi-automatically through CI (specify the base branch, click a button and have CI spawn the jobs). For GitHub Actions we could use https://www.actionspanel.app/ (I don't know if there is an equivalent solution for CircleCI)
3. CI could monitor jobs for failures and respawn them when the job failure is transient. We could also have a manual CI action which clones a job given its identifier.

[2opremio]

Some other things which we could do:

1. Automatically push the verify-range image when modifying it
2. Automatically test the verify-range image in CI with a small range

[ire-and-curses]

More improvements: https://github.com/stellar/pipelines/pull/202