Achieving a representative sample of marked animals: a spatial approach to evaluating post‐capture randomization