Implement more resampling schemes

[AdrienCorenflos]

1. Implement Srinivasan Sampling Process (SSP) resampling. Ref: https://arxiv.org/abs/1707.01845
2. Implement chopthin resampling. Ref: https://arxiv.org/abs/1502.07532. Note this cannot (directly?) be used with waste-free SMC
3. Implement Metropolis (and rejection perhaps) resampling. Ref: Parallel resampling in the particle filter. Note, the Metropolis resampler produces a biased sample.
4. Verify if non-exchangeable resampler can be used withing residual resampling, if so make it work.
5. Implement OT resampling.

[rlouf]

I guess that for now SSP would make the most sense?

[AdrienCorenflos]

I'd say SSP and metropolis yeah. Chopthin would require returning modified weights after resampling instead of assuming they are uniform.