Add EKS operator for commands in existing Pods

[AlejandroMorgante]

Add EksPodExecOperator so Dags can execute commands in running containers of existing Amazon EKS Pods without Airflow creating, restarting, or deleting them.

This supports long-lived workloads managed by another platform, where Airflow should own the command execution but not the Pod lifecycle. Reusing an already-running Pod also avoids paying its startup cost for every task.

The operator uses the AWS connection to generate a temporary EKS kubeconfig, then delegates command execution, log streaming, exit-code handling, and bounded optional XCom output to KubernetesPodExecOperator. This keeps EKS authentication in the Amazon provider while reusing the Kubernetes behavior introduced in #71244.

The change includes unit coverage, documentation, and an EKS system-test path that creates and deletes the external Pod outside the operator.

The live test also exposed an existing shell-portability issue in the EKS kubeconfig credential parser. The regression fix protects both EksPodOperator and EksPodExecOperator when credential discovery writes an informational line before the token output.

Testing:

• breeze run --skip-image-upgrade-check pytest providers/amazon/tests/unit/amazon/aws/operators/test_eks.py -q --tb=short (66 passed)

• breeze run --skip-image-upgrade-check pytest providers/amazon/tests/unit/amazon/aws/hooks/test_eks.py::TestEksHook -q (7 passed)

• breeze run --skip-image-upgrade-check mypy providers/amazon/src/airflow/providers/amazon/aws/hooks/eks.py providers/amazon/src/airflow/providers/amazon/aws/operators/eks.py providers/amazon/tests/system/amazon/aws/example_eks_with_nodegroups.py (no issues)

• System-test import and collection (1 test collected)

• Live AWS end-to-end system test using --forward-credentials: create the EKS cluster and nodegroup, create an externally managed Pod, execute and validate the command output, delete the Pod, and tear down the AWS resources (1 passed in 15:37)

• Apache Magpie self-review (pr-management-code-review, dry-run): completed with no code-level findings.

---

Was generative AI tooling used to co-author this PR?

• Yes — Codex (GPT-5)

Generated-by: Codex (GPT-5) following the guidelines

[AlejandroMorgante]

I also validated the operator with a real end-to-end run against an existing Pod in an EKS cluster. The task connected to the running Pod, executed the command, streamed its logs, and completed successfully in approximately 20 seconds. The Pod remained running and its lifecycle was not managed by Airflow.

This confirms the intended use case without provisioning a new Pod for each command.

[AlejandroMorgante]

@SameerMesiah97, could you please review this when you have a chance? This is the follow-up to #71244, adding the EKS integration for executing commands in existing Pods. Thank you!