[Feature] Support Pipeline Parallism in torchrun SPMD offline inference for V1 (#17827)

Signed-off-by: Lucia Fang <fanglu@fb.com>

vllm/v1/worker/gpu_worker.py

@@ -275,13 +275,13 @@ class Worker(WorkerBase):
 
         output = self.model_runner.execute_model(scheduler_output,
                                                  intermediate_tensors)
-
-        if not get_pp_group().is_last_rank:
+        parallel_config = self.vllm_config.parallel_config
+        if parallel_config.distributed_executor_backend != "external_launcher" \
+            and not get_pp_group().is_last_rank:
             assert isinstance(output, IntermediateTensors)
             get_pp_group().send_tensor_dict(output.tensors,
                                             all_gather_group=get_tp_group())
             return None
-
         assert isinstance(output, ModelRunnerOutput)
         return output if self.is_driver_worker else None