ImageEn arc rotation sample
==========================

Built with ImageEn 15.2.1 (15.2.1.15.0270).

imageen-arc-rotation-sample.ien   the layers, TImageEnIO.SaveToFileIEN
imageen-arc-rotation-sample.png   what this build draws for them, at 1:1

Every cell holds TWO polyline layers at the same place.

  GREEN  wide, underneath: the same path with the arc
         already converted to cubic Beziers BEFORE the
         layer was rotated.
  BLACK  narrow, on top: the path carrying an iepaArcTo
         segment.

So the black line should stay inside the green band. At
Rotate = 0 it does, over the whole path -- which is the
control: it says the two paths ARE the same curve, and
that nothing but the rotation is being compared.

An affine image of a cubic is the cubic through the transformed
control points, so the green layer is exact at every angle and is
what the black one is meant to look like. At Rotate = 0 the two
coincide. From 15 degrees on they do not, and at 33 the black arc
bulges the other way.

Row 1 is a bare arc, row 2 the same arc as the last of four segments.
Both rows are in the sample because the fit factors are computed from
the whole point cloud, so the two arcs are squeezed by different
amounts and fail differently.
