[Feature] CatBoost MultiQuantile support

[daidahao]

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 #2967.

Summary

I asked Codex for a quick summary:

• Adds native multi-quantile support for CatBoostModel by using CatBoost’s MultiQuantile loss (loss_function="MultiQuantile:alpha=...") instead of training one model per quantile.
• Updates quantile handling in the sklearn likelihood path so prediction works both when:
  • quantiles are stored in _model_container (old behaviour for non-catboost models), and
  • quantiles come directly from one native model output (new CatBoost behaviour).

Practical impact

• CatBoostModel quantile training should be faster/simpler (single native multi-quantile model path where supported).
• Existing quantile workflows remain supported for other models via fallback container-based logic.

Usage

import time

from darts.datasets import AirPassengersDataset
from darts.models import CatBoostModel

series = AirPassengersDataset().load()
train, val = series[:-36], series[-36:]
model = CatBoostModel(
    lags=12,
    output_chunk_length=6,
    likelihood="quantile",
    quantiles=[0.1, 0.5, 0.9],
)

start_time = time.perf_counter()
model.fit(series)
print(f"Training time: {time.perf_counter() - start_time:.2f} seconds")
"""
CatBoost without MultiQuantile:
Training time: 4.21 seconds

CatBoost with MultiQuantile:
Training time: 2.51 seconds
"""

Other Information

[codecov]

Codecov Report

❌ Patch coverage is 94.11765% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 95.67%. Comparing base (fdce8f2) to head (7f21c20).

Files with missing lines	Patch %	Lines
darts/models/forecasting/catboost_model.py	87.50%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3032      +/-   ##
==========================================
- Coverage   95.75%   95.67%   -0.08%     
==========================================
  Files         158      158              
  Lines       17181    17186       +5     
==========================================
- Hits        16451    16443       -8     
- Misses        730      743      +13     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[dennisbader]

Thanks a lot @daidahao 🚀 I'll be out of office next week but will have a look once I'm back :)

[daidahao]

@dennisbader
Any update on this? Thanks!