z-axis does not quite have zero at the top

[JeffreyEarly]

Somehow this bug got re-introduced during the refactor, but the top point on the z-axis of non-constant stratification isn't exactly zero.