One row per species within a named zone (subregion_key, programarea_key
or ecoregion_key), aggregated with pct_covered weighting.
Value
tibble, same shape as species_for_cells()
Details
Computed live from zone_cell + model_cell + taxon rather than read from
a precomputed table: v7 shipped a zone_taxon table, but v8 does not build
one, which left the app's species table broken. Measured on the v8 database,
the largest zone (subregion_key = "USA", ~349k cells, ~10k species) takes
~5 s, so precomputation is not required.