feat(llm): 每个 LLM 配置支持流式/非流式切换(默认流式)

依赖 norma v0.3.0 的 Provider.Complete。为每个 llm_profile 增加 streaming
开关:开(默认)走流式 SSE;关走真·非流式(stream:false、一次性返回完整
JSON),可绕开部分网关糟糕的 SSE 实现(空帧、思考字段丢帧),代价是失去
运行中的实时进度与实时 token 计数。

- agent.Config.Stream → agentcore.Options.NonStreaming = !Stream;worker /
  planner / mainagent / chat / goals 经 SetNonStreaming resolver 注入,任务
  路由按当前激活 profile 动态取值
- llmpool.Pool / llmrec.Recorder / taskLLMRuntime 三个 Provider 包装均实现
  Complete(分别带故障转移 / 录制 / 同 provider 重试与额度 failover),使
  输出、token 统计、操作统计、LLM 录制四条链路在非流式下保持兼容
- db: llm_profiles 新增 streaming 列 + ALTER 补旧库(默认 true,旧配置无感)
- 系统 → LLM 配置表单增加「流式输出」开关

server/server_mgmt.go

@@ -1736,10 +1736,14 @@ func (s *Server) pgSaveProfile(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 	// APIKey has json:"-" on db.LLMProfile so it is never leaked to the UI on
-	// read; accept it here via a sibling field for create/update.
+	// read; accept it here via a sibling field for create/update. Streaming is a
+	// *bool shadowing the embedded json:"streaming": an absent field must default
+	// to streaming (true), which a plain bool zero value (false = non-streaming)
+	// would get wrong — legacy/partial clients that never send it stay streaming.
 	var body struct {
 		db.LLMProfile
-		APIKey string `json:"api_key"`
+		APIKey    string `json:"api_key"`
+		Streaming *bool  `json:"streaming"`
 	}
 	if err := decode(r, &body); err != nil {
 		writeErr(w, 400, err.Error())
@@ -1747,6 +1751,7 @@ func (s *Server) pgSaveProfile(w http.ResponseWriter, r *http.Request) {
 	}
 	p := body.LLMProfile
 	p.APIKey = body.APIKey
+	p.Streaming = body.Streaming == nil || *body.Streaming
 	id, err := pg.SaveProfile(&p)
 	if err != nil {
 		writeErr(w, 500, err.Error())