Fix zero total population bug in segregationcalc

[tylersimko]

Hi, great package! Big fan.

I think there's a bug in the internal segregationcalc function when there's 0 population in a district. Edge case for redistricting of course, but this comes up for one of my applications.

For example:

library(redistmetrics)

data(nh)
data(nh_m)

# hypothetical 3rd NH district with 0 population
nh$r_2020[nh$vap == 0] <- 3

seg_dissim(plans = nh$r_2020,
           shp = nh,
           group_pop = vap_hisp,
           total_pop = vap)

[1] NaN NaN NaN

Looks like a potential divide by 0 bug here, so I pushed a fix to src/segregation.cpp to add 0 to dissimilarity in those cases.

> seg_dissim(plans = nh$r_2020,
+            shp = nh,
+            group_pop = vap_hisp,
+            total_pop = vap)
[1] 0.01085658 0.01085658 0.01085658

[christopherkenny]

Thanks Dr. @tylersimko, looks good to me. On to Dr. @CoryMcCartan

[CoryMcCartan]

Refactored for speed & readability.

All tests pass