LIN QMA polygons don't line up with statistical areas

[quantifish]

aa <- nz_fisheries_general_statistical_areas %>%
  dplyr::select(Statistica) %>%
  st_transform(crs = proj_nzsf()) %>%
  # st_snap(x = ., y = ., tolerance = 0.0001) %>%
  # st_simplify(dTolerance = 1) %>%
  st_union(by_feature = TRUE) %>%
  mutate(area = case_when(
    Statistica %in% c(401:412, "049", "050", "051", "052") ~ "a",
    Statistica %in% 601:625 ~ "b",
    TRUE ~ as.character("c")
  )) %>%
  group_by(area) %>%
  summarize(geometry = st_union(geometry))

ggplot() +
  geom_sf(data = aa, aes(fill = area)) +
  plot_qma(qma = "LIN", fill = "transparent") +
  # plot_statistical_areas(area = "stat area", fill = "transparent") +
  # plot_coast(resolution = "med", fill = "forestgreen", colour = "black", size = 0.3) +
  annotation_north_arrow(location = "tl", which_north = "true", style = north_arrow_nautical)

Also, there are some polygon slivers in the north island which I couldn't get rid of with snap or simplify.

[quantifish]

r-spatial/sf#547