hotfix: autoscaler queue_depth NULL depends_on blinded scaler

queue_depth turned a NULL depends_on into [None], and
ready_tasks then required None to be in completed_ids, so
such tasks never counted as ready. Worker pick_next was
unaffected (_task_from_row maps NULL to []), which is why
tasks still ran while the scaler always saw depth 0 and
never scaled up (proven live: depth probe with 1 queued
task returned 0). NULL now means no dependencies, with a
regression test.

cloud/autoscaler.py

@@ -65,7 +65,15 @@ def queue_depth(db_path, executor=None):
             deps = row["depends_on"]
         except (KeyError, IndexError, TypeError):
             deps = []
-        if isinstance(deps, str):
+        if deps is None:
+            # depends_on 为 NULL（裸 SQL 插的任务、或老数据）表示
+            # 无依赖，绝不能变成 [None]——那样 ready_tasks 会判定
+            # 它依赖一个永远不存在的 None，任务永远不算就绪，
+            # scaler 也就永远看不到深度（2026-10-26 实测教训：
+            # worker 的 _task_from_row 对 NULL 处理正确所以能
+            # 正常认领，只有 scaler 的深度计算被这个 bug 致盲）。
+            deps = []
+        elif isinstance(deps, str):
             try:
                 deps = json.loads(deps) if deps else []
             except ValueError: