[SUGGESTION] Add more declick techniques

[MarcoRavich]

Hi there,
I've noticed that declick uses "Sliding-window local RMS" but I believe that would be interesting to have other algos as listed in the @rajmic's A Survey and an Extensive Evaluation of Popular Audio Declipping Methods.

From VI. CONCLUSION AND PERSPECTIVES chapter:

methods based on social shrinkage, nonnegative matrix factorization, weighting the transform coefficients and, last but not least, the SPADE seem to yield results that make them preferred choices; in most conditions, they scored significantly better than the majority of other methods in terms of the perceptual metrics.

The relative git provides codes too, anyway here are some - maybe helpful - implementations:

• @andryr's spade-declipping;
• @pierreaguie's C-OMP-for-Audio-Declipping;
• @LucasRr's Consistent dictionary learning for signal declipping;
• @MarwanMashra's / @space928's Audio Declipping with Social Sparsity;
• @Dimdp's audio-declipper (wavelet);
• @cocuh's declipnet (wavenet);
• @pierreaguie's not-MIWAE for Audio Declipping.

Hope that inspires and helps.