Metric calculation is bogus

precision calculation is currently taking predictions for several as of dates, and calculating precision across all of them together, resulting in bogus results. need to look at how to do it for each as of date separately and then aggregate or something more reasonable.

Start by tracing the precision calculation and how predictions from several as-of dates are combined. Determine a reasonable per-date calculation and aggregation approach, then verify that precision is no longer computed by pooling all dates together.

Written by the indexing model from the issue text.