Allow constant pmins and pmaxs?

[tjthurman]

No description provided.

[tjthurman]

Would allow BAHZ to replicate the fixed at 0/1 models from HZAR. You can more-or-less do this already: use a super-tight uniform prior aroudn the desired value (e.g., 0.7999 to 0.80001 to do 0.8, or 0.9999 to 1 to do 1). Stan does not have a "constant" probability distribution that can be used as a prior, so for now that seems like the best way. It could also probably be done with optional parameters (see #67), but that would require some major changes in how the priors get implemented.

[tjthurman]

Add a check to make sure bounds for uniform priors aren't equal