feat(agent): 移除control_node实例化，新增系统节点命名与人设管理前端

当前阶段只保留regulatory+consciousness两个系统节点，control_node代码保留但不再实例化。
系统节点新增display_name字段支持自定义显示名称，前端新增人设管理Tab支持模板CRUD。

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

main.py

@@ -41,7 +41,6 @@ from kilostar.core.global_state_machine import GlobalStateMachine
 from kilostar.core.global_workflow_manager import GlobalWorkflowManager
 from kilostar.core.individual.regulatory_node import RegulatoryNode
 from kilostar.core.individual.consciousness_node import ConsciousnessNode
-from kilostar.core.individual.control_node import ControlNode
 
 if KILOSTAR_MODE != "standalone":
     import ray
@@ -73,12 +72,13 @@ async def start_standalone():
     consciousness_node = ConsciousnessNode()
     register_standalone("consciousness_node", consciousness_node)
 
-    control_node = ControlNode()
-    register_standalone("control_node", control_node)
-
-    worker_cluster = WorkerCluster()
+    worker_cluster = WorkerCluster(node_type="cpu")
     await worker_cluster.start()
     register_standalone("worker_cluster", worker_cluster)
+    # 单机模式三个标签共用同一实例
+    register_standalone("worker_cluster_cpu", worker_cluster)
+    register_standalone("worker_cluster_core", worker_cluster)
+    register_standalone("worker_cluster_gpu", worker_cluster)
 
     print(f"✅ KiloStar 单机模式启动完成，监听 0.0.0.0:8000")
 
@@ -129,15 +129,22 @@ async def start_distributed():
 
     RegulatoryNode.options(name="regulatory_node").remote()
     ConsciousnessNode.options(name="consciousness_node").remote()
-    ControlNode.options(name="control_node").remote()
 
     try:
-        WorkerCluster.options(
-            name="worker_cluster", lifetime="detached"
-        ).remote()
-        print("✅ WorkerCluster 已成功启动并注册！")
-    except ValueError:
-        print("WorkerCluster 已经存在。")
+        for node_type in ("cpu", "core", "gpu"):
+            actor_name = f"worker_cluster_{node_type}"
+            resource_key = f"kilostar_node_{node_type}"
+            try:
+                WorkerCluster.options(
+                    name=actor_name,
+                    lifetime="detached",
+                    resources={resource_key: 1},
+                ).remote(node_type=node_type)
+                print(f"✅ WorkerCluster[{node_type}] 已成功启动并注册！")
+            except ValueError:
+                print(f"WorkerCluster[{node_type}] 已经存在。")
+    except Exception as e:
+        print(f"WorkerCluster 启动失败: {e}")
 
     print("正在等待 GlobalWorkflowManager 初始化与恢复工作流...")
     try: