Bring in binned cluster counts likelihood

[itrharrison]

The current clusters likelihood is for unbinned clusters. Following Planck (and others) @eunseongleee has implemented a binned cluster counts likelihood following the Cash statistics. This is now at the stage of having been well tested in situ, so that a fiducial likelihood value can be calculated.

This work is currently on the binned_cluster branch, and contains quite a lot of development cruft and overlap with other likelihoods.

We should start a new branch dev-binned_clusters and start bringing things which are strictly necessary across from the binned_clusters branch.

The first things to implement would be a dummy likelihood which is a subclass of CashLikelihood (#54 ) which returns the fiducial likelihood value, and a test which... tests this against the fiducial value.