At times, our API deployments fail due to a fluke. It could be valuable to automatically re-run these deployments up to some set number of times, then stop. We'd like to explore the following questions:

1. Is it possible to automatically re-run a failed deploy action?
2. How difficult is it to code this into our existing GitHub Actions?
3. Can we set a specific failure threshold, after which point the action does not run again?
4. What are any trade-offs here?