[Core] Use sha256 bytes instead of BlockHash to reduce GC overhead (#23673)

Signed-off-by: linzebing <linzebing1995@gmail.com>

examples/online_serving/kv_events_subscriber.py

@@ -6,6 +6,8 @@ import msgspec
 import zmq
 from msgspec.msgpack import Decoder
 
+from vllm.v1.core.kv_cache_utils import BlockHash
+
 
 #
 # Types copied from vllm.distributed.kv_events
@@ -22,8 +24,8 @@ class KVCacheEvent(
 
 
 class BlockStored(KVCacheEvent):
-    block_hashes: list[int]
-    parent_block_hash: Optional[int]
+    block_hashes: list[BlockHash]
+    parent_block_hash: Optional[BlockHash]
     token_ids: list[int]
     block_size: int
     lora_id: Optional[int]
@@ -31,7 +33,7 @@ class BlockStored(KVCacheEvent):
 
 
 class BlockRemoved(KVCacheEvent):
-    block_hashes: list[int]
+    block_hashes: list[BlockHash]
     medium: Optional[str]