ELUC: Predictor robojudge

[ofrancon]

Add a "robojudge" that evaluates how well the predictors perform and how they compare.
Different metrics can be computed and compared, like mean absolute error (MAE) or MAE / hectar, mean square error, etc. Mean rank is another useful metric: rank the predictors for each country, and compute their mean rank on aggregate.
It should be possible to compare the predictors at an aggregated level, and also at a country level.

See https://phase1.xprize.evolution.ml/ for an example of robojudge that was used for the Pandemic Resilience XPRIZE challenge.
See https://github.com/cognizant-ai-labs/covid-xprize/blob/master/predictor_robojudge.ipynb for a notebook that was used to compare the predictors