KubernetesExecutor Worker Configuration in Airflow 3.0

### Apache Airflow Provider(s) cncf-kubernetes ### Versions of Apache Airflow Providers With Airflow 3.0 and the new Task SDK architecture, the airflow worker command no longer exists, but there's insufficient documentation on how to properly configure worker images for KubernetesExecutor environments using the CNCF Kubernetes provider. ## Environment - Airflow Version: 3.0.6 - Provider Version: apache-airflow-providers-cncf-kubernetes>=8.1.0 - Python Version: 3.12 - Executor: KubernetesExecutor - Deployment: Kubernetes - Operating System: Linux (Docker containers) ## Background We are migrating from Airflow 2.x to 3.0 and need clarification on the proper way to configure worker containers when using KubernetesExecutor from the CNCF Kubernetes provider, especially given the new Task SDK and API server architecture changes in Airflow 3.0. ## Breaking Change: airflow worker Command Removed in Airflow 3.0 The airflow worker command that existed in Airflow 2.x has been completely removed in Airflow 3.0 due to the new Task SDK architecture (AIP-72). In the previous architecture, workers were long-running processes that connected directly to the metadata database. With Airflow 3.0, the architecture has shifted to a client-server model where workers communicate through a dedicated API server instead of direct database access. For KubernetesExecutor specifically, this means worker pods no longer run persistent worker processes - instead, the KubernetesExecutor directly injects task execution commands (like airflow tasks test) into ephemeral pods that execute individual tasks and terminate. This architectural change improves security by removing direct database access from workers and enables better isolation, but requires updating worker image configurations to remove any references to the obsolete airflow worker command. “Is apache-airflow-providers-cncf-kubernetes compatible? Do you have an example you could share?” ### Apache Airflow version 3.0.6 ### Operating System Li…