AutoMLForecast .fit() method does not work when id_col != "unique_id" or target_col != 'y' due to defaults

[bryanwhiting]

What happened + What you expected to happen

Tried fitting AutoMLForecast with a series that has different id_col and getting an error saying "unique_id" not found. Problem is that id_col isn't being passed into smape

https://github.com/Nixtla/mlforecast/blob/main/mlforecast/auto.py#L515

workaround for now is to just name my column as unique_id.

Versions / Dependencies

mlforecast.__version__ = '0.13.5'

Reproduction script

from mlforecast.auto import AutoMLForecast, AutoLightGBM

from statsforecast.utils import generate_series

series = generate_series(n_series=10, seed=1)

fcst = AutoMLForecast(
    models = [AutoLightGBM()],
    freq='D',
    season_length=7
)
fcst.fit(series, num_samples=2, n_windows=1, h=7)


series.columns = ['org', 'time' , 'target']
fcst.fit(
    series,
    id_col='org',
    time_col='time',
    target_col='target',
    num_samples=2,
    n_windows=1,
    h=7
)

Issue Severity

Medium: It is a significant difficulty but I can work around it.