chore: import upstream snapshot with attribution

doc/source/serve/doc_code/async_inference_autoscaling.py

@@ -0,0 +1,40 @@
+# __basic_example_begin__
+from ray import serve
+from ray.serve.config import AutoscalingConfig, AutoscalingPolicy
+from ray.serve.schema import CeleryAdapterConfig, TaskProcessorConfig
+from ray.serve.task_consumer import task_consumer, task_handler
+
+processor_config = TaskProcessorConfig(
+    queue_name="my_queue",
+    adapter_config=CeleryAdapterConfig(
+        broker_url="redis://localhost:6379/0",
+        backend_url="redis://localhost:6379/1",
+    ),
+)
+
+
+@serve.deployment(
+    max_ongoing_requests=5,
+    autoscaling_config=AutoscalingConfig(
+        min_replicas=1,
+        max_replicas=10,
+        target_ongoing_requests=2,
+        policy=AutoscalingPolicy(
+            policy_function="ray.serve.async_inference_autoscaling_policy:AsyncInferenceAutoscalingPolicy",
+            policy_kwargs={
+                "broker_url": "redis://localhost:6379/0",
+                "queue_name": "my_queue",
+            },
+        ),
+    ),
+)
+@task_consumer(task_processor_config=processor_config)
+class MyConsumer:
+    @task_handler(name="process")
+    def process(self, data):
+        return f"processed: {data}"
+
+
+app = MyConsumer.bind()
+serve.run(app)
+# __basic_example_end__