Every result written by this project carries one of these. It exists because
fleet's own snapshots had the provenance inverted: the 25-minute comparison
recorded the malariasimulation version, the R version, the replicate count
and a digest of the scenarios, while the seven-hour site-file run – the one
nobody can repeat – recorded only a date and a fleet version. The tier that
cannot be re-run is the tier that most needs to say what made it.