[FEAT] Transfer learning with Conformal Prediction

[nasaul]

Closes #456

• Code refactor
• Re-calibration functionality. This needs data for the calibration.
• Weighted Conformal prediction This controls for feature shifting.
• Normalized Conformal Prediction scaled per error in each serie. This needs data for calibration.
• Normalized Conformal Prediction scaled per serie. This controls for the scale of the serie.
• Weighted Conformal prediction + Normalized Conformal Prediction = Zero shot transfer learning intervals
• Experiment
• Update docs

[elessar-for-github]

Changes made by this pull request:

mlforecast/conformal_prediction.py
[Error processing diff]

mlforecast/forecast.py
[Error processing diff]

mlforecast/utils.py
[Error processing diff]

tests/test_forecast.py
[Error processing diff]

[codspeed-hq]

Merging this PR will not alter performance

✅ 12 untouched benchmarks

---

_{Comparing feat/transfer_cp (42e8665) with main (b41fc7d)}