Sebastian.p/generalized rff

[SebastianPopescu]

The purpose of this PR is to to support sampling with models

• that can have SeparateIndependent and SharedIndependent kernels & inducing variables.
• with Heteroskedastic likelihoods (so multiple GP heads)

Main changes:

• creation of feature_decomposition_kernels folder. Idea was to structure it just like in GPflow (i.e. multioutput subfolder). Having a separate folder for this type of kernels will prove to be better suited if we are planning on including in the future some other papers such as .. [1] Solin, Arno, and Simo Särkkä. "Hilbert space methods for reduced-rank Gaussian process regression." Statistics and Computing (2020). .. [2] Borovitskiy, Viacheslav, et al. "Matérn Gaussian processes on Riemannian manifolds." In Advances in Neural Information Processing Systems (2020).
• in gpflux.layers.basis_functions.fourier_features I have added the multioutput version
• in gpflux.sampling I have added the multioutput version

[sebastianober]

It seems like there may be a lot of added files from personal runs of the models that we won't want in GPflux, e.g., docs/notebooks/logs, cos/notebooks/ckpts, and docs/notebooks/checkpoint. Could you verify this?

[sebastianober]

I think we also don't typically do formatting checks on notebooks, although I don't see any need to revert this

[hstojic]

I have cleaned the PR, now ready for review!