Normalize uid argument to set_prognoses

[daniel-klein]

set_prognoses is either

def set_prognoses(self, sim, target_uids, source_uids=None)

or

def set_prognoses(self, sim, uids, source_uids=None)

Let's pick one and be consistent.

[cliffckerr]

For the specific question -- my vote is for uids.

However, I don't especially like the name set_prognoses -- I feel this is maybe a holdover from Covasim where we pre-computed the prognoses for the entire disease course, but we won't necessarily be doing that here. In particular, what I see the set_prognoses function doing is:

• Appending to the infection log
• Modifying state values (e.g. self.susceptible[uids] = False)
• In some cases, then actually setting prognoses (e.g. self.ti_primary[uids] = sim.ti + rr(self.pars.dur_exposed.rvs(uids) / sim.dt)

I wonder if we should rename make_new_cases to choose_cases, and set_prognoses to make_cases or something like that?

I also don't love "cases". For infectious diseases, "case" sometimes means only a diagnosed infection. I propose that we use the term infection in Infection and its subclasses, and do not define this method for Disease, since it may or may not be applicable. For example, I could imagine an Obesity module wouldn't involve a make_new_cases method that would flip agents from one state to another, but would instead track continuous variables, and over a certain threshold agents would be defined as being a "case".

[cliffckerr]

Closed by #527