feat(worker): 暂停/重跑续接先前 LLM 对话，不再从头开始

Worker 的 Execute 在创建 session 后尝试 Resume 先前的 transcript，
如果存在则续接对话而非重建全新会话，避免浪费已积累的 LLM 上下文。

Claude-Session: https://claude.ai/code/session_014TfZWeVGhjXqZD3vcj44mz

agent/worker.go

@@ -360,6 +360,18 @@ func (w *Worker) Execute(ctx context.Context, name string, taskID int64, as *db.
 	s := agentcore.NewSession(opts)
 	defer s.Close() // release the session's background-task manager (temp dir + processes)
 
+	// Resume prior conversation if this intent was paused/blocked/exhausted and is
+	// being re-run. The transcript ID is deterministic per intent, so if a prior
+	// session exists the worker continues from where it left off instead of
+	// restarting from scratch.
+	if w.tx != nil {
+		_ = s.Resume(opts.SessionID)
+		if len(s.Messages()) > 0 {
+			seedUnlockFromHistory(s.Messages(), def.UnlockSkill)
+			input = "继续执行。"
+		}
+	}
+
 	// Budgets + settlement are owned by the SDK (MaxTurns/MaxDuration + Settlement):
 	// on hit it runs a wrap-up turn and finishes with ReasonMaxTurns/ReasonTimeout.
 	// We only add a HARD ctx backstop past the soft wall-clock budget, for the rare