[BUG] Ray executor raises ValueError: Tileable object must be executed first before fetch

[fyrestone]

Describe the bug A clear and concise description of what the bug is.

mars/learn/tests/test_wrappers.py:25 (test_parallel_post_fit_basic)
setup = <mars.deploy.oscar.session.SyncSession object at 0x295346820>

    def test_parallel_post_fit_basic(setup):
        raw_x, raw_y = make_classification(n_samples=1000)
        X, y = mt.tensor(raw_x, chunk_size=100), mt.tensor(raw_y, chunk_size=100)
        clf = ParallelPostFit(GradientBoostingClassifier())
>       clf.fit(X, y)

mars/learn/tests/test_wrappers.py:30: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
mars/learn/wrappers.py:173: in fit
    return self._make_fit("fit")(X, y=y, **kwargs)
mars/learn/wrappers.py:148: in _fit
    mr.spawn(_wrap, args=(self.estimator, method, X, y), kwargs=kwargs)
mars/core/entity/executable.py:144: in execute
    return execute(self, session=session, **kw)
mars/deploy/oscar/session.py:1888: in execute
    return session.execute(
mars/deploy/oscar/session.py:1682: in execute
    execution_info: ExecutionInfo = fut.result(
../../.pyenv/versions/3.8.13/lib/python3.8/concurrent/futures/_base.py:444: in result
    return self.__get_result()
../../.pyenv/versions/3.8.13/lib/python3.8/concurrent/futures/_base.py:389: in __get_result
    raise self._exception
mars/deploy/oscar/session.py:1868: in _execute
    await execution_info
../../.pyenv/versions/3.8.13/lib/python3.8/asyncio/tasks.py:695: in _wrap_awaitable
    return (yield from awaitable.__await__())
mars/deploy/oscar/session.py:105: in wait
    return await self._aio_task
mars/deploy/oscar/session.py:953: in _run_in_background
    raise task_result.error.with_traceback(task_result.traceback)
mars/services/task/supervisor/processor.py:372: in run
    await self._process_stage_chunk_graph(*stage_args)
mars/services/task/supervisor/processor.py:250: in _process_stage_chunk_graph
    chunk_to_result = await self._executor.execute_subtask_graph(
mars/services/task/execution/ray/executor.py:553: in execute_subtask_graph
    meta_list = await asyncio.gather(*output_meta_object_refs)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

awaitable = ObjectRef(359ec6ce30d3ca2dffffffffffffffffffffffff0100000001000000)

    @types.coroutine
    def _wrap_awaitable(awaitable):
        """Helper for asyncio.ensure_future().
    
        Wraps awaitable (an object with __await__) into a coroutine
        that will later be wrapped in a Task by ensure_future().
        """
>       return (yield from awaitable.__await__())
E       ray.exceptions.RayTaskError(ValueError): ray::execute_subtask() (pid=70399, ip=127.0.0.1)
E         File "/home/admin/mars/mars/services/task/execution/ray/executor.py", line 187, in execute_subtask
E           execute(context, chunk.op)
E         File "/home/admin/mars/mars/core/operand/core.py", line 491, in execute
E           result = executor(results, op)
E         File "/home/admin/mars/mars/core/custom_log.py", line 94, in wrap
E           return func(cls, ctx, op)
E         File "/home/admin/mars/mars/utils.py", line 1160, in wrapped
E           return func(cls, ctx, op)
E         File "/home/admin/mars/mars/remote/core.py", line 154, in execute
E           result = function(*function_args, **function_kwargs)
E         File "/home/admin/mars/mars/learn/wrappers.py", line 34, in _wrap
E           X = X.fetch() if isinstance(X, ENTITY_TYPE) else X
E         File "/home/admin/mars/mars/core/entity/executable.py", line 164, in fetch
E           return self._fetch(session=session, **kw)
E         File "/home/admin/mars/mars/core/entity/executable.py", line 160, in _fetch
E           self._check_session(session, "fetch")
E         File "/home/admin/mars/mars/core/entity/executable.py", line 152, in _check_session
E           raise ValueError(
E       ValueError: Tileable object f3164084e1a24b42818a147a8ccc7be5_0 must be executed first before fetch

../../.pyenv/versions/3.8.13/lib/python3.8/asyncio/tasks.py:695: RayTaskError(ValueError)

To Reproduce To help us reproducing this bug, please provide information below:

1. Your Python version
2. The version of Mars you use
3. Versions of crucial packages, such as numpy, scipy and pandas
4. Full stack of the error.
5. Minimized code to reproduce the error.

Expected behavior A clear and concise description of what you expected to happen.

Additional context Add any other context about the problem here.