Annotation name to aggregate across completed test runs.
OptionaldirectionOptimization direction; defaults to "maximize". "maximize" treats a
higher mean as better (clears when >= threshold); "minimize" treats a
lower mean as better (clears when <= threshold) — use it for cost,
latency, or error-rate annotations.
The bar the mean score must clear, compared in direction. Boolean scores
average as 1 (true) / 0 (false).
Gate the suite on the mean score: the average across all runs must clear
threshold(compared indirection).