It seems like it takes half an hour or an hour to update scores for a single IFP. Is this a bottleneck for deployment of CultivateLabs software in a scaled up live forecasting situation? Or is it an artifact of running everything on a single AWS t3.nano instance?