Run
data class RunResponseTableData(var runId: Int = -1, var responseName: String = "", var average: Double = 0.0, var variance: Double? = null, var count: Double = 0.0) : DbTableData(source)
One row per (run, response): the cell best's estimate for every response of the problem, the objective included.
This is what makes a selection replayable. tblRun preserves the best point's INPUTS, which is enough to re-simulate but not to re-select: ranking a candidate needs its average, variance and count per response. With those stored, an alternative confirmation rule can be applied to a finished study offline, and only the finalists it chooses need fresh simulation.
A single-observation estimate legitimately has no variance. It is stored as null, since SQLite stores NaN as NULL and a NOT NULL column would reject the whole batch the row belongs to.