Keep _process_priors() and _unload_params() in CLVModel

[ColtAllen]

Currently, _process_priors() and _unload_params() are being defined in every CLV model, but the only differences are the specific prior/param names. If we add both methods to the base CLVModel as generic loop operators, and a list attribute in the child model classes containing the param names, it would eliminate a lot of redundant code when adding new models. #133 partially resolves _process_priors().

Also, in each child model class every predictive method calls _unload_params() and clv.utils.to_xarray(). It may be worthwhile to write a decorator in CLVModel for predictive methods that handles this automatically, but I need to look into this more.

[ColtAllen]

I started working on this, but I'm not sure if these code changes can accommodate covariates, so I'll return to this after progress has been made on #134.

[ColtAllen]

I'm currently adding support for covariates to ParetoNBDModel, which creates nuances to _unload_params that are not agnostic across all models. _process_priors is also in the base class now, so I'm closing this.