[PD] Fix dynamic port support and MLA buffer for Mooncake (#5415)

Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
Co-authored-by: ybyang <ybyang7@iflytek.com>

python/sglang/srt/disaggregation/base/conn.py

@@ -5,6 +5,7 @@ import numpy as np
 import numpy.typing as npt
 
 from sglang.srt.disaggregation.utils import DisaggregationMode
+from sglang.srt.server_args import ServerArgs
 
 
 class KVArgs:
@@ -16,6 +17,7 @@ class KVArgs:
     aux_data_lens: list[int]
     aux_item_lens: list[int]
     ib_device: str
+    gpu_id: int
 
 
 class KVPoll:
@@ -30,7 +32,12 @@ class BaseKVManager(ABC):
     """Base class for managing transfers states"""
 
     @abstractmethod
-    def __init__(self, args: KVArgs, disaggregation_mode: DisaggregationMode): ...
+    def __init__(
+        self,
+        args: KVArgs,
+        disaggregation_mode: DisaggregationMode,
+        server_args: ServerArgs,
+    ): ...
 
 
 class BaseKVSender(ABC):