Graceful Shutdown and Lifecycle Management in ECS Fargate

Act as a Node.js/Python Backend Architect. Detail the critical programmatic implementation of graceful container shutdowns for an application executing within the ephemeral ECS Fargate environment. You must configure the `stopTimeout` parameter within the ECS task definition strictly to [STOP_TIMEOUT] seconds (overriding the 30s default). Write the precise application-level [SIGTERM_HANDLER_LOGIC] required to intercept the OS signal, stop accepting new HTTP requests, flush active database queues, and exit cleanly with code 0. Mathematically correlate this container lifecycle behavior directly with the AWS ALB [TARGET_GROUP_ARN] configuration, explicitly proving why the ALB's [DEREGISTRATION_DELAY] must be perfectly synced with the task timeout to absolutely prevent 502 Bad Gateway errors during scaling events.