Get Exponential models working

[drbenvincent]

First brought in with a commit under #131, but still need to:

• [x] get a sensible prior for k
• [x] confirm convergence works well

[drbenvincent]

These models work fine. Testing has shown that convergence can be a bit problematic for the Kirby demo dataset. But this is a data issue. Testing on other (unpublished) data from my lab using an alternative (not Kirby method) shows these Exponential discounting models can produce good, stable, converged inferences. People should update priors to suit their situation