Start a minimal `fitting` module.

[marina-ricci]

After discussion within the group we converge on starting a fitting module as part of CLMM that would contains :

• method to derive the mass posterior for individual clusters
• method to derive the mass (+ concentration) posterior for a stack of clusters

We could start by a backward model approach (e.g. likelihood of model given data) and then allow for a forward model approach (e.g. likelihood of data given model). As a start we can start using NumCosmo in place of Firecrown, and storing data vectors as sacc files.