add future X_df features to predict

[JLiekenbrock]

Description

In this PR MLForecast Core is updated so that using predict with direct approach will correctly use the future external regressors.

Checklist:

• This PR has a meaningful title and a clear description.
• The tests pass
• All linting tasks pass.
• The notebooks are clean.

The linting tasks fail, but not because of things i have changed

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[jmoralez]

Thanks, but the fit part is also using the exogenous at the first step, so that has to be fixed first.

[JLiekenbrock]

For the fit part i find it most straight forward to just add another branch to the fit method
and move the loop from fit_models one level up and just shift the exogenous regressors there.
One could also consider to add X_df as parameter to the fit function but X_df is anyways already in the data.

The fitted branch is not updated yet, When moving the loop over the models outside of the _compute_fitted_values then cross_validation would need to be updated too.

[codspeed-hq]

CodSpeed Performance Report

Merging #504 will not alter performance

_{Comparing JLiekenbrock:bug_-core]-Direct-Forecasts-Incorrect-Treatment-of-Exogenous-Variables-bug (258433f) with main (f674b45)}

🎉 Hooray! pytest-codspeed just leveled up to 4.0.0!

A heads-up, this is a breaking change and it might affect your current performance baseline a bit. But here's the exciting part - it's packed with new, cool features and promises improved result stability 🥳!
Curious about what's new? Visit our releases page to delve into all the awesome details about this new version.

Summary

✅ 12 untouched benchmarks