feat: Handle missing ECS deployment gracefully when replaced

[Bankamah]

This PR improves the robustness of the ECS deployment polling script. Previously, if a deployment was replaced by a manual or external redeploy while the GitHub Action was polling, the script would fail or exit with unclear messaging. Now, if the deployment ID is no longer found (indicating a replacement), the script exits gracefully with a clear message.

[iBotPeaches]

Interesting. I guess I'm confused that a deployment stepped over by an external one should still exist. I'm itching to see the raw response since I'm not understanding what is missing or blank.

[Bankamah]

When an external or manual redeploy occurs in ECS, the original deployment can be removed from the service’s deployments array almost immediately. In this case, the deployment ID the script is polling for is no longer present in the response from aws ecs describe-services. That’s why $DEPLOYMENT ends up blank.

Would you like for me to add a debug print?

[iBotPeaches]

Would you like for me to add a debug print?

I think if we mess with an internal project (template possibly). Lets re-deploy Alpha, then force a deployment in AWS to replicate the error. Once we see we are replicating the error in the polling, we can make the API call manually in CURL to get body. Can you setup the tools to run that test to get a good real sample?

[Bankamah]

I can give it a shot.