Consider how users could extend formula used to include other population stratifications (sex, workplace, etc)

[njtierney]

One approach could be that users change the formula e.g., add y ~ s(workplace), and then we parse out the new stratification term and then manage all the data manipulation etc.

It might simpler perhaps though to add something a stratification argument and then the user passes the bare variable name (bare name, as this indicates that this is part of the data).

e.g.,

fit_single_contact_model(
  contact_data,
  population,
  symmetrical = TRUE,
  # this is the new argument
  stratification = workplace_name,
  school_demographics = NULL,
  work_demographics = NULL
)

And the populate the stratification argument across the relevant parts of the package code