Priors on stellar parameters

Priors on stellar parameters

The problem: I suspect that a common use case might resemble the following scenario: A user would like to force the value of a parameter to either a single value, or else a distribution- for example a mean and variance. Perhaps she has astro-seismic log g, or metallicity measured from a companion. Or she might wish to constrain more than one stellar property or calibration parameter. The existing framework currently assumes uniform priors for stellar parameters over the library interval. There exists a "fix_logg" mechanism to fix the log g to a singular value, but it's labeled as a "durty hack".

Suggested solution:
Add a mechanism for picking among an array of priors, or a user configurable prior. The default prior would remain as stated in Czekala et al. 2015, but through some flag, a user could specify a normal distribution, or whatever.

The issue names no files or tests. Start by locating the current uniform-prior handling and the existing fix_logg mechanism, then determine how configurable priors would fit the inference framework. Done should include a user-selectable prior while preserving the current default behavior.