Completeness and Purity kernels with wrong equation and suggestion implementation

[eduardojsbarroso]

New feature requests and enhancements
I suggest we implement the completeness and purity kernels in firecrown/models/cluster/kernel.py with free parameters that can be sampled by cosmosis.

Problem description
First of all, there is an error in the current kernel implementation. Current implementation is using logm/logmc whereas it should be m/mc, as per Aguena, 2016. However I also propose that we implement these functions as we did in mass_proxy.py, where the parameters will be set by cosmosis instead of hard coded. Even if we do not sample over these parameters, they will be different for different input data (i.e. galaxy cluster catalogs obtained from different cluster finders). Also, I think it would be helpful if we can sample over the parameters. We can find a fit from cosmological simulations and then I suggest a gaussian prior centered on these fit parameters to give some freedom for the fit.

Proposed solution
Change Completeness and Purity classes in firecrown/models/cluster/kernel.py to be updatable objects with 4 parameters each.