Fix incorrect usage of os.getenv in register_kv_caches method

Author: leideng

ucm/integration/vllm/ucm_connector.py

@@ -232,7 +232,7 @@ def _generate_storage_backends(
             return [path for path in storage_backends.split(":") if path]
 
     def register_kv_caches(self, kv_caches: dict[str, torch.Tensor]):
-        if os.getenv["VLLM_HASH_ATTENTION"] == "1":
+        if os.getenv("VLLM_HASH_ATTENTION", "0") == "1":
             for layer_name, value in kv_caches.items():
                 kv_cache, k_hash = value
                 self.kv_caches[layer_name] = kv_cache