Syncing lineages between model and eval data

[afmagee42]

I'm looking at the pair of model.csv and eval.csv for a 2022-01-01 forecast date, and the evaluation data has one more lineage than the model data, namely 22C. We need to make sure that the set of lineages is consistent across modeling and evaluation.

Part of the awkwardness here is hindsight, in that we have said 22C is a lineage we want to model for all three of the forecast dates in retrospective-forecasting/config even though it emerges after the first calibration period.

I think the generally correct thing to do is to make sure such lineages get labeled as "other" but there may be the possibility that we know we want to model a lineage but just don't happen to see it? (Something which has attracted attention elsewhere, but not been imported yet, for instance.)