Fix ReplicaContext serialization error in DPRankAssigner

Remove unused replica_ctx parameter from DPRankAssigner.register() to
fix serialization error when passing ReplicaContext across actor
boundaries.

Root cause: The addition of _handle_registration_callback field to
ReplicaContext in PR ray-project#58345

Fixes ray-project#58474 and ray-project#58475

Signed-off-by: Nikhil Ghosh <nikhil@anyscale.com>

Author: nrghosh

python/ray/llm/_internal/serve/serving_patterns/data_parallel/dp_rank_assigner.py

@@ -55,13 +55,12 @@ def __init__(self, dp_size: int, dp_size_per_node: Optional[int] = None):
             )
 
     async def register(
-        self, replica_ctx: "serve.context.ReplicaContext", node_id: Optional[str] = None
+        self, node_id: Optional[str] = None
     ):
         """
         Register a replica and assign a rank to it.
 
         Args:
-            replica_ctx: The replica context.
             node_id: The node id of the replica.
 
         Returns:

python/ray/llm/_internal/serve/serving_patterns/data_parallel/dp_server.py

@@ -24,9 +24,8 @@ class DPServer(LLMServer):
     async def __init__(self, llm_config: LLMConfig, dp_rank_assigner: DeploymentHandle):
         self.dp_rank_assigner = dp_rank_assigner
 
-        replica_ctx = serve.get_replica_context()
         node_id = get_runtime_context().get_node_id()
-        self.dp_rank = await self.dp_rank_assigner.register.remote(replica_ctx, node_id)
+        self.dp_rank = await self.dp_rank_assigner.register.remote(node_id)
 
         logger.info(f"DP rank {self.dp_rank} registered with rank assigner")