Fix error when handing over trainer_kwargs to predict_depency()

[jurgispods]

Description

This PR fixes an error when handing over trainer_kwargs to predict_dependency(), as in the following example:

model.predict_dependency(ds, "feature", np.linspace(0, 30, 30), mode="dataframe",  trainer_kwargs={ "accelerator": "cpu"})

Overwriting trainer_kwargs might be necessary when using a model trained on GPU. The issue comes from the kwargs dict handed over to predict_dependency() being overwritten in predict(), which will cause it to contain a PredictCallback(return_index=True, ...) in the kwargs["trainer_kwargs"]["callbacks"] list, even in the second iteration where return_index should be False, causing a subsequent error

File ~/Library/Caches/pypoetry/virtualenvs/discovery-lab-rTrwb4a5-py3.10/lib/python3.10/site-packages/pytorch_forecasting/models/base_model.py:1491, in BaseModel.predict_dependency(self, data, variable, values, mode, target, show_progress_bar, **kwargs)
   1488 data.reset_overwrite_values()  # reset overwrite values to avoid side-effect
   1490 # results to one tensor
-> 1491 results = torch.stack(results, dim=0)
   1493 # convert results to requested output format
   1494 if mode == "series":

TypeError: expected Tensor as element 1 in argument 0, but got list

This PR fixes this by copying the user-specified kwargs in every iteration. No tests have been added, as this is a very minor change.

Checklist

• Linked issues (if existing)
• Amended changelog for large changes (and added myself there as contributor)
• Added/modified tests
• 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!

[codecov-commenter]

Codecov Report

Patch coverage: 100.00% and project coverage change: +0.08% 🎉

Comparison is base (f0ead9e) 90.12% compared to head (8aa3d3d) 90.21%.

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

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1366      +/-   ##
==========================================
+ Coverage   90.12%   90.21%   +0.08%     
==========================================
  Files          30       30              
  Lines        4711     4712       +1     
==========================================
+ Hits         4246     4251       +5     
+ Misses        465      461       -4     

Flag	Coverage Δ
cpu	90.21% <100.00%> (+0.08%)	⬆️
pytest	90.21% <100.00%> (+0.08%)	⬆️

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

Files Changed	Coverage Δ
pytorch_forecasting/models/base_model.py	88.19% <100.00%> (+0.01%)	⬆️

... and 1 file with indirect coverage changes

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