Describir: Extending Multi-Output Methods for Long-Term Aboveground Biomass Time Series Forecasting Using Convolutional Neural Networks