`castSeedDataQueensNeighbourhood` cannot handle situation with a single infection

[bryce-carson]

# A tibble: 4 × 8
  Location      lat   lon InitialVaccinated InitialExposed InitialInfections InitialRecovered InitialDead
  <chr>       <dbl> <dbl>             <dbl>          <dbl>             <dbl>            <dbl>       <dbl>
1 Balzers      47.1  9.50                 0              0                 0                0           0
2 Schaan       47.2  9.51                 0              0                 0                0           0
3 Triesenberg  47.1  9.54                 0              0                 0                0           0
4 Vaduz        47.1  9.52                 0              0                 1                0           0

This seed data results in the function stopping because somewhere near Vaduz a cell will have a negative value in Susceptible, which is a violation.

[bryce-carson]

This situation will arise when a cell in the Moore neighbourhood has a value of zero, as $I_{ij} = 0 - x$ is negative in that cell (castSeedDataQueensNeighbourhood() is called with neighbourhood.order = 1).

[bryce-carson]

@ashokkrish, there are two ways to handle this:

1. Deal with individuals in a clever manner, such as using conditions to check that there are sufficient people to distribute, and follow a particular or random pattern of distribution when there isn't, or
2. Use the density of people (rather than a count of actual individuals) in the cells; i.e. it is okay for "half a person" to be infected, etc.

In the first solution, separate branches of statements would handle cases wherein there are sufficient individuals to distribute over the cells, insufficient individuals but more than one, and only one. The second and third cases would be handled by seeding one individual in the central cell, and in the second case proceeding to seed the remaining individuals in some appropriate, unbiased manner.