Tasks towards better non-Gaussian uncertainty handling

[MartaVanin]

1. gradlogpdfs and heslogpdfs should always return Floats; Exponential returns an Int
2. add docs (check docstrings!)
3. add pluto notebook in examples with minimal se (3bus case) with different distributions
4. add "PV" case from the paper
5. add automatic support for Laplacian distribution in (r)wlav case