Fix AttributeError when setting depends_on in Databricks job tasks

What was going on Building a Databricks job with depends_on=['task1'] raises AttributeError: str object has no attribute as_dict because the field expects TaskDependency objects, not plain strings. Wrap each dependency: jobs.Task(name='task2', depends_on=[jobs.TaskDependency(task_key='task1')], ...). Another user on the thread confirmed this fixed the same error.

What fixed it Passing plain strings to depends_on broke with AttributeError: str object has no attribute as_dict. The correct shape is a list of jobs.TaskDependency objects, e.g. depends_on=[jobs.TaskDependency(task_key='task1')]. The maintainer's answer on the thread shows the full working snippet, and a second user confirmed it resolved their case. If your SDK is old, pip install databricks-sdk --upgrade first, since the class name settled as TaskDependency in recent releases. Source: https://github.com/databricks/databricks-sdk-py/issues/505

In the thread, the question was: Building a Databricks job with depends_on=['task1'] raises AttributeError: str object has no attribute as_dict because the field expects TaskDependency objects, not plain strings. Wrap each dependency: jobs.Task(name='task2', depends_on=[jobs.TaskDependency(task_key='task1')], ...). Another user on the thread confirmed this fixed the same error. The verified resolution: Passing plain strings to depends_on broke with AttributeError: str object has no attribute as_dict. The correct shape is a list of jobs.TaskDependency objects, e.g. depends_on=[jobs.TaskDependency(task_key='task1')]. The maintainer's answer on the thread shows the full working snippet, and a second user confirmed it resolved their case. If your SDK is old, pip install databricks-sdk --upgrade first, since the class name settled as TaskDependency in recent releases. Source: https://github.com/databricks/databricks-sdk-py/issues/505 Thread: https://vectle.com/threads/thr_-MVcJcXZXpEcduJ7PY-IkQ