[1] in my own words:
* starting from plot-level data with one row per plot,
* i need to build a household-season dataset with mean yield, total nitrogen,
* number of plots, and an indicator for any irrigated plot.

[2] inputs i need:
* - dataset: plot_dataset.dta
* - ids: hh_id_merge, plot_id_merge
* - outcome variables: harvest_kg, nitrogen_kg, total_labor_days, plot_area_GPS, irrigated

[3] outputs:
* - one row per hh_id_merge-season
* - variables: hh_id_merge, season, mean_yield_kg, total_fert, total_labor
*              n_plots, any_irrigated
* - save as $export/hh_yield.dta

[4] clarification questions:
* 1. should we include all seasons or only main season?
* 2. what should we do with plots that have missing inputs or plot_area_GPS?
* 3. should we drop households with only one plot?