Error: Weekly Naive model used on series < 7 samples

[capelastegui]

On #195 , we introduced auto detection of appropriate naive models. Unfortunately, this breaks for series without a full week worth of samples - a weekly naive model is used instead of the correct regular naive model.

[capelastegui]

Solution - In forecast_models.py:

• create new function, _can_use_season_weekly(), based on line 1949:

    use_season_weekly = (
        # Need more than a full week
        (max_date_delta > pd.Timedelta(min_periods * 7, unit='d')) &
        # Need at least daily samples
        (min_date_delta <= pd.Timedelta(1, unit='d'))
    )

• Replace the naive model selection code from:

        l_model_naive = [forecast_models.model_snaive_wday] if freq == 'D' \
            else [forecast_models.model_naive]

to:

l_model_naive = [forecast_models.model_snaive_wday] if _use_season_weekly(df_y) \
            else [forecast_models.model_naive]