combinef function to constrain bottom level predictions with external predictions in gts

[amelinevallet]

Hello,
I have a population dataset with 3 grouping variables (Age, Gender and Urban/Rural Setting).
I would like to generate bottom level predictions ( for Age * Gender * UrbanRuralSetting) and to make them match with higher level predictions (Age*Gender) that we have from external sources (outputs of a model).
I am discovering the hts package, but I think this is something I can do with the combinef function. Is it correct?
Is there a better way to proceed with the functions provided by the package ? I was initially thinking to use an arima model with the higher levels as external predictors (following this tutorial : https://robjhyndman.com/hyndsight/hts-with-regressors/), but I have read that external predictors should be other explanatory variables, and not predictions. Do you confirm this point? I'll be happy to provide a reproductible example once I am sure of the function I should use.
Best regards,
Améline Vallet