Assignment columns with 0s fail

[pjrule]

This issue was reported by @gabeschoenbach and can be resolved by incrementing each value in the assignment column by 1. Assignment values are used as array indices in some places—we normally assume that assignments start from 1 and are in the range [1, n_districts]. However, this is not universally true—some assignments may start with 0 and may be non-consecutive. Let's talk about the cleanest way to deal with this. One easy option is to simply enforce our assumption in a preprocessing step and shift the burden to the user, but I'm not sure what the UX would be here.