Switch to Era5t data

[miltminz]

Hello everybody, in this PR the following changes

• The modelling phase, especially the predict, has been changed to perform on ERA5T data instead of ERA5Land. The advantage is that we can query more recent data to predict (up to day -5 from present)
• New models trained on some ERA5T variables have been uploaded to the release.
• The predict module has been adapted and it now lets predict with the XGBoost model as well. The add_lags function is applied only to model columns where lags are used, which speeds it up a bit.
• Some minor adaptations in unitary tests (which take now less time as well)

[codecov]

Codecov Report

Merging #29 (4796754) into master (8c840df) will decrease coverage by 0.51%.
The diff coverage is 73.68%.

@@            Coverage Diff             @@
##           master      #29      +/-   ##
==========================================
- Coverage   95.04%   94.53%   -0.52%     
==========================================
  Files          15       15              
  Lines         686      695       +9     
==========================================
+ Hits          652      657       +5     
- Misses         34       38       +4     

Flag	Coverage Δ
unittests	94.53% <73.68%> (-0.52%)	⬇️

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

Impacted Files	Coverage Δ
pyro_risks/models/predict.py	84.61% <68.75%> (-8.94%)	⬇️
pyro_risks/datasets/era_fwi_viirs.py	93.61% <100.00%> (+0.13%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8c840df...4796754. Read the comment docs.

[frgfm]

Thanks for the PR! "Speed it up a bit", don't be so modest haha, going from 22min+ unittest to 15min is quite an advantage 🙏