Extend model to allow arbitrary number of covariates

[sempwn]

problem statement

This will require a change in how model is currently structured. Currently when running the model,

est_naloxone(d)

the function expects that d contains the column region which provides a unique key for each factor in the model (either region or site type, or a combination of the two). Instead we want d to contain an arbitrary number of columns each representing its own covariate.

design approach

Obvious choice would be to follow in the vain of brms and other packages and use the formula approach to specify the model covariates e.g.

est_naloxone(~region + site_type, data = d)

This allows us to use the R formula machinery and in particular the model.matrix function to generate a design matrix and also allows for incorporating interaction via,

X <- model.matrix(~region + site_type + region:site_type, data = d)

The default formula can be set to ~region to ensure backwards compatibility.