The new values can’t fit in the predefined keys for the subdivision type selection.
Running
import pycountry
for t in sorted({sub.type.lower() for sub in pycountry.subdivisions}): print(t)
for pycountry 20.7.3 and 22.1.11 will yield the different type sets with the following additions
administrative precinct
autonomous city in north africa
capital
city municipality
decentralized regional entity
district municipality
district with special status
districts under republic administration
federal capital territory
free municipal consortium
group of islands (20 inhabited islands)
land
metropolitan administration
metropolitan city
metropolitan collectivity with special status
overseas collectivity
overseas collectivity with special status
pakistan administered area
regional state
rural municipality
special administrative city
special self-governing city
special self-governing province
town
urban community
urban municipality
voivodship
ward
and modifications
chains (of islands) | chain (of islands)
federal territories | federal territory
island group | islands, groups of islands
overseas territorial collectivity | overseas territory
popularates | popularate
and many deletions.
So a debate is open on how tryton should keep up with this rather arbitrary and evolving data set.
Regarding the released 6.0 LTS, the subdivision type selection could be extended to the union of the different value sets that each pycountry version would yield.
Otherwise, the extra dependency trytond-country[data] should state to depend on pycountry==20.7.3 because pycountry does not abide by semver and any future or past release could potentially break the import script. Yet this would be quite troublesome for pip dependency resolution.