[ENH] Support tuning any model and extend LSTMModel in docs to support multi-target datasets

[svnv-svsv-jm]

Description

This PR fixes my own issues:

• Example on multiple targets (with proposed solution) #1440
• Tune anything (with solution) #1448

It allows to tune "any" model, generalizing the optimize_hypetparameters function from TemporalFusionTransformer model, and it also add a "better" LSTMModel than the one shown in the documentation, which I tried to use on a multi-target dataset without success. My version does work on multi-target datasets, thanks to an extension to AutoRegressiveBaseModel.

My version of AutoRegressiveBaseModel inherits from the original to override a couple of methods. I should've probably just edited the original, but I'd like to hear your feedback before doing that. I'm also sure there was a simpler way for AutoRegressiveBaseModel to work with multi-target data. When I tried to just run the Documentation's example of LSTMModel but on a multi-target dummy dataset, it just did not work out of the box.

Checklist

• Linked issues: https://github.com/jdb78/pytorch-forecasting/issues/1440 , https://github.com/jdb78/pytorch-forecasting/issues/1448
• Amended changelog for large changes (and added myself there as contributor)
• Added/modified tests: tests/test_models/test_tuning.py
• [Yes] Used pre-commit hooks when committing to ensure that code is compliant with hooks. Install hooks with pre-commit install.
  To run hooks independent of commit, execute pre-commit run --all-files

Make sure to have fun coding!

[svnv-svsv-jm]

What's blocking this?

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[svnv-svsv-jm]

I don't understand why the docs test fails (I haven't changed anything there) and why the Black test fails if I ran black and on my machine black --check . returns:

>> black --check .
All done! ✨ 🍰 ✨
54 files would be left unchanged.

[codecov-commenter]

Codecov Report

Attention: 35 lines in your changes are missing coverage. Please review.

Comparison is base (b3fcf86) 90.19% compared to head (652a4c1) 89.96%.
Report is 8 commits behind head on master.

Files	Patch %	Lines
pytorch_forecasting/models/_base_autoregressive.py	77.46%	16 Missing ⚠️
pytorch_forecasting/models/tuning.py	85.43%	15 Missing ⚠️
pytorch_forecasting/data/timeseries.py	0.00%	3 Missing ⚠️
pytorch_forecasting/models/lstm.py	98.83%	1 Missing ⚠️

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

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1449      +/-   ##
==========================================
- Coverage   90.19%   89.96%   -0.23%     
==========================================
  Files          30       33       +3     
  Lines        4724     4985     +261     
==========================================
+ Hits         4261     4485     +224     
- Misses        463      500      +37     

Flag	Coverage Δ
cpu	89.96% <86.79%> (-0.23%)	⬇️
pytest	89.96% <86.79%> (-0.23%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.