I found this excellent sweep vignette authored by @Matt Dancho, but I want to make sure that the process follows a top-bottom approach, i.e. that the forecast of subgroups would equal to the forecast of total.
To use Matt's example, if bike_sales_monthly and monthly_qty_by_cat2 tables were forecasts (not actuals), I want to makes sure that the summed monthly_qty_by_cat2$total.qty to year and month level will equal bike_sales_monthly$total.qty.
This is definitely supported in hts package but I can't use it easily with tidyverse and more importantly, hts/gts models don't give you forecast confidence intervals.