Improvement/Adds temporal_hidden_size_past and temporal_hidden_size_future hyperparams to TiDEModel

[eschibli]

Checklist before merging this PR:

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

Implements #2408

Summary

Adds hyperparameters controlling hidden layer size for the past covariate and future covariate feature encoders in TiDEModel. Previously these were controlled by hidden_size.

Other Information

Defaults were set to 64. In the original google-research implementation the default temporal_hidden_size_past was set to 64, and there was no future covariate encoder.

Alternatively, to preserve default behavior, defaults could be set to 128.

Other comments:

Oops I see my rebase failed horribly - should I drop this PR and do it over from a fresh branch?

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 50.00000% with 5 lines in your changes missing coverage. Please review.

Project coverage is 93.77%. Comparing base (2f71ce8) to head (b068996).

Files	Patch %	Lines
darts/models/forecasting/tide_model.py	50.00%	5 Missing :warning:

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2416      +/-   ##
==========================================
- Coverage   93.81%   93.77%   -0.05%     
==========================================
  Files         139      139              
  Lines       14694    14690       -4     
==========================================
- Hits        13785    13775      -10     
- Misses        909      915       +6     

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.