Rounding Behavior Change for Double

round(value) and round(value, scale) functions for Double now align with the IEEE 754 specification and may behave differently on boundary cases.

If you need the legacy behavior — which disregards Unit of Least Precision (ULP) arithmetic inaccuracies — use the new roundStrict(value) and roundStrict(value, scale) functions instead.