fix: Set identifier to esa-prod for continuity

[ceholden]

What I am changing

This PR updates the deploy step to point to the same stack that has been running (esa-prod, not prod). This change had happened as part of the switch to the ESA Copernicus services instead of IntHub. These changes weren't propagated back to "infrastructure as code" and the downloader stack wasn't redeployed using the IDENTIFIER=prod target, so this PR seeks to resolve this "drift"

I could see this going the other way (tear down "esa-prod" stack after getting "prod" to work) but having the cdk diff against existing stack seems safer to me than a full cutover without an easy way to compare the stacks. Additionally there are a number of secrets that "esa-prod" has already configured.

How I did it

Find/replace

How you can test it

Integration tests should still work and we'll have a fairly small cdk diff that mostly includes,

• Definition of Lambda assets is shorter in CDK v2 for the S3Key (no need to lookup refs from stack). Generally there's a few .Ref changes to be .Fn::Sub, which in other words means that the metadata is more directly specified
• Python 3.8 ~> 3.11 runtime
• A few StepFunction changes,
  • CDK v2 includes an extra AWS as defaults that retry will run on
  • The Principal in AssumeRolePolicyDocument for step function is now just states.amazonaws.com

[chuckwondo]

Reviewing this was rather arduous. Perhaps in the future you can break up such large changes into smaller pieces :-)

[sharkinsspatial]

@ceholden This approach is good with me. For historical context we needed to maintain both deployments in parallel previously while we were diagnosing ongoing problems with the CDE publication streams.