sktime: [BUG] AutoETS `predict_quantile` and `predict_interval` fail in sktime 0.17.2

Describe the bug

AutoETS predict_quantiles and predict_interval fail in sktime 0.17.2, These were working in 0.17.1. Originally observed when we tried to upgrade pycaret to use sktime 0.17.2 (see: https://github.com/pycaret/pycaret/issues/3523).

To Reproduce

!pip install sktime==0.17.2

from sktime.datasets import load_airline
from sktime.forecasting.ets import AutoETS

y = load_airline()
forecaster = AutoETS(auto=True, n_jobs=-1, sp=12)  
forecaster.fit(y)  

# Gives error
y_pred = forecaster.predict_quantiles(fh=[1,2,3])

# Also gives error
y_pred = forecaster.predict_interval(fh=[1,2,3])

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
[<ipython-input-4-b9df560b190c>](https://localhost:8080/#) in <cell line: 1>()
----> 1 y_pred = forecaster.predict_quantiles(fh=[1,2,3])

4 frames
[/usr/local/lib/python3.10/dist-packages/sktime/forecasting/base/_base.py](https://localhost:8080/#) in predict_quantiles(self, fh, X, alpha)
    559         # we call the ordinary _predict_quantiles if no looping/vectorization needed
    560         if not self._is_vectorized:
--> 561             quantiles = self._predict_quantiles(fh=fh, X=X_inner, alpha=alpha)
    562         else:
    563             # otherwise we call the vectorized version of predict_quantiles

[/usr/local/lib/python3.10/dist-packages/sktime/forecasting/base/_base.py](https://localhost:8080/#) in _predict_quantiles(self, fh, X, alpha)
   2077 
   2078                 # compute quantile forecasts corresponding to upper/lower
-> 2079                 pred_a = self._predict_interval(fh=fh, X=X, coverage=[coverage])
   2080                 pred_int = pd.concat([pred_int, pred_a], axis=1)
   2081 

[/usr/local/lib/python3.10/dist-packages/sktime/forecasting/base/adapters/_statsmodels.py](https://localhost:8080/#) in _predict_interval(self, fh, X, coverage)
    189         valid_indices = fh.to_absolute(self.cutoff).to_pandas()
    190 
--> 191         prediction_results = self._fitted_forecaster.get_prediction(
    192             start=start, end=end, exog=X
    193         )

[/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/exponential_smoothing/ets.py](https://localhost:8080/#) in get_prediction(self, start, end, dynamic, index, method, simulate_repetitions, **simulate_kwargs)
   2100         """
   2101         return PredictionResultsWrapper(
-> 2102             PredictionResults(
   2103                 self,
   2104                 start,

[/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/exponential_smoothing/ets.py](https://localhost:8080/#) in __init__(self, results, start, end, dynamic, index, method, simulate_repetitions, **simulate_kwargs)
   2296             if ndynamic:
   2297                 sim_results.append(
-> 2298                     results.simulate(
   2299                         ndynamic,
   2300                         anchor=anchor_dynamic,

TypeError: ETSResults.simulate() got an unexpected keyword argument 'exog'

Expected behavior This worked in sktime 0.17.1 so the behavior should have continued in 0.17.2

Versions

/usr/local/lib/python3.10/dist-packages/_distutils_hack/init.py:33: UserWarning: Setuptools is replacing distutils. warnings.warn(“Setuptools is replacing distutils.”)

System: python: 3.10.11 (main, Apr 5 2023, 14:15:10) [GCC 9.4.0] executable: /usr/bin/python3 machine: Linux-5.15.107±x86_64-with-glibc2.31

Python dependencies: pip: 23.1.2 sktime: 0.17.2 sklearn: 1.2.2 numpy: 1.22.4 scipy: 1.10.1 pandas: 1.5.3 matplotlib: 3.7.1 joblib: 1.2.0 statsmodels: 0.13.5 numba: 0.56.4 pmdarima: None tsfresh: None tensorflow: 2.12.0 tensorflow_probability: 0.19.0