Use NULL for onset-to-event and risk function arguments

[joshwlambert]

This PR addresses #147 by replacing the use of NA with NULL for turning off onset-to-event delays (e.g. onset_to_hosp or onset_to_death) or risks (e.g. hosp_risk or hosp_death_risk).

The default value for onset_to_recovery has been changed from NA to NULL in sim_linelist() and sim_outbreak(). The documentation of function argument has been updated, as has the unit tests and a vignette.

Internal checking functions have been updated to check for NULL values. The as_function() converts NULL to the same NA generating function, therefore the output line list still contains a column of NAs which theses processes are turned off.

[sbfnk]

Great!