ENH: add wavelet-based motion correction for fNIRS (motion_correct_wavelet)

[leonardozaggia]

Reference issue (if any)

None (new feature).

What does this implement/fix?

Adds motion_correct_wavelet (alias wavelet) to mne.preprocessing.nirs for wavelet-based motion correction of fNIRS data.

The function pads each channel to a power of 2, normalises by the estimated noise level (MAD of first-level detail coefficients), applies a stationary wavelet transform (SWT), zeroes out detail coefficients that are IQR-based outliers, and reconstructs the corrected signal. Specialises in spike removal.

Based on Homer3 v1.80.2 hmrR_MotionCorrectWavelet (Huppert et al. 2009) and Molavi & Dumont 2012. Adds new dependency: PyWavelets

Additional information

• Operates on optical density or hemoglobin data.
• iqr controls the outlier threshold; set to -1 to skip thresholding entirely.
• wavelet and level select the mother wavelet and decomposition depth.
• This is one of three related PRs adding Homer3-based motion correction to mne.preprocessing.nirs
  - PCA: ENH: add PCA-based motion correction for fNIRS (motion_correct_pca) #13691
  - Spline: ENH: add spline interpolation motion correction for fNIRS (motion_correct_spline) #13693
  - Wavelet: this PR

[welcome]

Hello! 👋 Thanks for opening your first pull request here! ❤️ We will try to get back to you soon. 🚴