model options for interventions

[sbfnk]

Thanks to the API refactor we now have sensible groupings of options which make the interface much clearer.

Now might be a good time to take a broader view of these and think about whether we want to re-group / change any of the options groupings before the next release, especially in the light of potential future generalisations as suggested in #68 and #114.

Some interrelated questions are:

1. How do we model future additional interventions, e.g. (1)

   scenario_sim(
     vaccination = vaccination(),
     quarantine = quarantine_opts(),
     testing = testing_opts(),
     pep = pop_opts()
   )

   or (2)

   scenario_sim(
     intervention = intervention_opts(
       vaccination = vaccination(),
       quarantine = quarantine_opts(),
       testing = testing_opts(),
       pep = pop_opts()
     )
   )

   (my initial preference: option (2))

2. Isolation is an intervention -- do we want isolation_opts() for related arguments (in whichever we form decide for (1))? This could take e.g. a delay_to_isolation (which could be from onset or positive test), isolated_offspring etc.

3. Does it make sense to have delays, offspring and event_probs, or do we want disease_opts() for everything that doesn't depend on interventions (incubation period, R0, R_asymp, proportion asympomatic, proportion of transmisison before symptoms)?