Interface with ducc0 noise generation and documentation improvements

[paganol]

Description

This PR introduces an interface with the ducc0 noise generation module, offering a significant performance boost for long timelines. It refactors the noise.py module to support multiple noise generation engines and physical models, while improving documentation and defaulting to the standard FFT-based approach for backward compatibility.

Key Changes

New Noise Engines & Models:

• Added support for engine="ducc", which uses ducc0.misc.OofaNoise for time-domain filtering. This is faster and more memory-efficient for long simulations.
• Added support for model="keshner" (native to DUCC) alongside the existing model="toast", which implements the model in toast.
• Refactored add_one_over_f_noise to dispatch between fft (default) and ducc engines.

API Updates:

• add_noise_to_observations and underlying functions now accept engine and model arguments.
• Defaults: The defaults remain engine="fft" and model="toast" to ensure backward compatibility.
• Validation: Added safety checks to prevent mismatched engine/model combinations (e.g., raising a ValueError if engine="ducc" is used with model="toast").

[github-actions]

Coverage report

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
litebird_sim
hwp_diff_emiss.py					80
noise.py					21-27, 36-42, 72-85, 215-229, 396
simulations.py
Project Total

This report was generated by python-coverage-comment-action