Fix/mc dropout

[dennisbader]

Checklist before merging this PR:

• Mentioned all issues that this PR fixes or addresses.
• Summarized the updates of this PR under Summary.
• Added an entry under Unreleased in the Changelog.

Fixes #2300.

Summary

• fixes Monte Carlo Dropout for all TorchForecastingModels. MC dropout training mode was not getting activated properly in the different model training/val/pred/.. stages.
• add MonteCarloDropout to all TorchForecastingModels that were not using them before.

Additional Information

Below an example that it works properly now.

import numpy as np
import matplotlib.pyplot as plt

from darts import concatenate
from darts.models import TCNModel, TSMixerModel
from darts.datasets import AirPassengersDataset

series = AirPassengersDataset().load().astype(np.float32)
preds = []

dropouts = [0., 0.5, 0.99]
for dropout in dropouts:
    model = TCNModel(13, 12, dropout=dropout, random_state=42)
    model.fit(series, epochs=100)
    preds.append(
        concatenate(
            model.historical_forecasts(
                series=series,
                retrain=False,
                forecast_horizon=12,
                stride=12,
                last_points_only=False,
            ),
            axis=0
        )
    )

series.plot()
for pred, dropout in zip(preds, dropouts):
    pred.plot(label=f"d={dropout}")
plt.show()

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 94.01%. Comparing base (883e35e) to head (bb8f15c).

❗ Current head bb8f15c differs from pull request most recent head b95337b. Consider uploading reports for the commit b95337b to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2312      +/-   ##
==========================================
- Coverage   94.01%   94.01%   -0.01%     
==========================================
  Files         138      138              
  Lines       14145    14138       -7     
==========================================
- Hits        13299    13292       -7     
  Misses        846      846              

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