[FEAT] Baseline models insample prediction intervals

[AzulGarza]

Is your feature request related to a problem? Please describe. For StatsForecast.forecast_fitted_values to work properly with levels, the baseline models (HistoricAverage, Naive, SeasonalNaive, RandomWalkWithDrift) need to have the in-sample prediction intervals implemented.

Additional context The in-sample prediction intervals returned by the methods [Model].forecast and [Model].predict_in_sample should be a dict with keys fitted-lo-{lv}, fitted-hi-{lv} for each lv in levels.