Fix/optimized historical forecast with component specific lags

[madtoinou]

Summary

Optimized historical forecasts with regression models would raise an exception when using component-specific lags with different number of lags because create_lagged_prediction_data() was calling model.lags.get() instead of the dedicated method model._get_lags() which properly accounts for the component-specific lags.

Other Information

Code snippet to reproduce the bug:

from darts.models import LinearRegressionModel
from darts.datasets import AirPassengersDataset

# future_cov components: darts_enc_fc_cyc_month_sin and darts_enc_fc_cyc_month_cos
model = LinearRegressionModel(
    lags=3,
    lags_future_covariates={'default_lags':[-4, -2],
                            'darts_enc_fc_cyc_month_sin':[-1]
                           },
    add_encoders={
        "cyclic": {"future": ["month"]},
    }
)

series = AirPassengersDataset().load()
model.fit(series)
model.historical_forecasts(series=series, retrain=False)

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Files	Coverage Δ
darts/models/forecasting/regression_model.py	96.50% <100.00%> (ø)
...orical_forecasts/optimized_historical_forecasts.py	97.56% <ø> (ø)

... and 6 files with indirect coverage changes

📢 Thoughts on this report? Let us know!.

[dennisbader]

Looks great, thanks for fixing this so quickly @madtoinou 🚀

I added a test, ready to be merged now 💯