Enable async scheduling

Signed-off-by: Jeffrey Wang <jeffreywang@anyscale.com>

vllm/config/vllm.py

@@ -687,7 +687,7 @@ def __post_init__(self):
             "mp",
             "uni",
             "external_launcher",
-        )
+        ) or (executor_backend == "ray" and envs.VLLM_USE_RAY_V2_EXECUTOR_BACKEND)
 
         if self.scheduler_config.async_scheduling:
             # Async scheduling explicitly enabled, hard fail any incompatibilities.