Infinite patient capacity

[glwagner]

The main purpose of this PR is to remove the field .patient_capacity from HealthService. The HealthService now has an infinite capacity. It also removes .waiting_list.

A few other changes are:

• Sets are used rather than lists for patients. This obviates duplicate entries and simplifies filtering operations.
• When patients are discharged, edges are not added to current hospital patients. This has no effect on the current model, but will in the case that resistant people can transition to susceptible (immunity has finite lifetime). This is important for some diseases and may be important for COVID-19. Because this is a simple change, we include it in this PR.
• We make sure not to add edges to ContactSimulator between discharged and current patients, and not to remove edges from ContactSimulator between new patients and previous patients (which do not exist because they were removed when the previous patient was admitted). This change is intended to makes the code's function more explicit and clear.

We have also changed printing for EpidemicSimulator and HealthService. Running examples/simple_epidemic.py now produces output like

                               *** Day: 2.500

[ Patient manifest ]          Admitted: 25237
                            Discharged: (3282, 'R')
                               Current: 8993, 5730, 25610, 29931, 4653, 18191, 8272, 25237, 4278, 7640, 20249, 7867

[ Status report ]          Susceptible: 28199
                               Exposed: 1136
                              Infected: 413
                          Hospitalized: 13
                             Resistant: 237
                              Deceased: 2
             Current possible contacts: 299739
               Current active contacts: 4497

[ Wall times ]    Hosp. administration: 0.0024 s,
                    Contact simulation: 1.4633 s,
                    Kinetic simulation: 3.1975 s

every static_contact_interval.

[odunbar]

regarding the set in what context were we getting duplicate entries? Was it when we search for hospital seeking nodes?

The other changes are fine - our model doesn't yet involve R->S as you note, so i am happy with the edges not being replaced.

[glwagner]

regarding the set in what context were we getting duplicate entries? Was it when we search for hospital seeking nodes?

We were not getting duplicate entries.

The point is that sets let us write

subset = large_group - group_to_ignore

which is simple. Sets do not allow duplicate entries, either. This is a downside if you expect to get duplicate entries. Since we don't expect duplicate entries, we can use sets.

The other changes are fine - our model doesn't yet involve R->S as you note, so i am happy with the edges not being replaced.

The point of the change is so that supporting this transition is possible (and also to make the function of the code more explicit).