5% was a guesstimate. here are the boundingboxes:

App.ActiveDocument.CurvedArray.Shape.Edges[1].BoundBox
BoundBox (-50, 20, -50, 50, 20, 50)
App.ActiveDocument.CurvedSegment.Shape.Edges[0].BoundBox
BoundBox (-49.9999, 22, -49.2467, 50, 22, 49.2459)

in this case the ribs are too short by 1.5mm for a diameter of 100mm - or 1.5%
the issue also happens when using 2 hullcurves

when not using a hullcurve, the size is correct - so the issue doesn't seem to be the shape interpolation of the ribs (which is trivial in this case) but the scaling to the hullcurve(s)

found the issue. scaleByBoundbox() computes the scale factor based on relative sizes of the bounding boxes. The bounding box of the hullcurve is calculated correctly. However the bounding box of the to be rescaled shape - in case of the CurvedArray is taken from the base shape - which is almost always correct (unless the base shape has been created by script)
while for a CurvedSegment, the shape is newly generated (half way between the end shapes) but has not yet been displayed.
FreeCAD only tesselates shapes when they are shown in the GUI. prior to tesselation the BoundBox is a rough estimate and can be too large by a few percent.

The following workaround fixes the issue:

--- a/CurvedSegment.py
+++ b/CurvedSegment.py
@@ -153,7 +153,8 @@ class CurvedSegmentWorker:
             normal = CurvedShapes.vectorMiddle(fp.NormalShape1, fp.NormalShape2, d)
             #Draft.makeLine(ribs[i].BoundBox.Center, ribs[i].BoundBox.Center + normal)
             bbox = CurvedShapes.boundbox_from_intersect(fp.Hullcurves, ribs[i].BoundBox.Center, normal, self.doScaleXYZ)
-            if bbox:              
+            if bbox:
+                ribs[i].tessellate(0.01) # needed for correct bounding box of newly created shapes
                 ribs[i] = CurvedShapes.scaleByBoundbox(ribs[i], bbox, self.doScaleXYZsum, copy=False)

see also https://forum.freecad.org/viewtopic.php?f=22&t=38402

Thank you for your contribution to fix this