Maybe changing the text structure import schema to something that's compatible with the internal locations data is the right way. So instead of a nested import structure, we'd expect a list of locations that have IDs made up by the user for an initial import (the downloaded template should illustrate this!) and the actual internal IDs if there already are locations in the DB for the target text. This way we could update location labels and aliases through an import later on.
Maybe changing the text structure import schema to something that's compatible with the internal locations data is the right way. So instead of a nested import structure, we'd expect a list of locations that have IDs made up by the user for an initial import (the downloaded template should illustrate this!) and the actual internal IDs if there already are locations in the DB for the target text. This way we could update location labels and aliases through an import later on.