chore: make cleanup a bit safer (#3346)

I have observed a lot of occasions where tests are being disturbed because
resources are being deleted out from under them; either buckets or roles are
being deletd while they are still being deployed.

- List all resources in CloudFormation, and avoid hard-deleting any resource
  that is still managed by a CloudFormation stack.
- In this PR, extend the notion of "staleness" (older than X hours) to
  stacks and apps, so we don't delete stacks and apps that might potentially
  still have tests running on them.
- Increase the staleness period a little, because I've seen cleanup be too
  eager.
- Put some time-traceability into the cleanup session's name, so we can
  start debugging a little easier than with just a random number.

Author: rix0rrr

packages/amplify-e2e-tests/package.json

@@ -41,6 +41,7 @@
     "@aws-sdk/client-amplify": "^3.812.0",
     "@aws-sdk/client-sts": "^3.812.0",
     "@aws-sdk/client-organizations": "^3.812.0",
+    "@smithy/util-retry": "^4.1.2",
     "amplify-category-api-e2e-core": "5.0.9",
     "aws-amplify": "^4.2.8",
     "aws-appsync": "^4.1.1",