Structure of the config file

[emhuff]

A dictionary containing: cosmology:

• an astropy cosmology object
• a list of parameters to vary, each element of which is a dictionary with the fields: name, value, lower, upper

mean systematics:

• a name of a function which returns a mean covariance
• a list of parameters to vary, same as above

systematics covariance:

• one name for a function which returns a covariance matrix substructure (identity, block diagonal, band diagonal, fully correlated, etc.)
• a list of parameters for each function, same as above

structure of the data

• a vector of redshifts at which to make the measurements
• a string that tells us which measurement is being made.

[bnord]

@reneehlozek @bfalck Do we want any of these added to the config file?

[bfalck]

for pendulum we are letting all "cosmology" and nuisance parameters vary... is there a reason to restrict params or range?

also this gives me an idea of how to include systematics functions in the config: if we pre-define a set of functions that determine the true systematics array, we can specify the function name in the config. it's not flexible but I don't think we can have a function as a value in the config dictionary otherwise.