The current sloppy bounding box is inconsistent for x and y (x<=width, y<height) and does not explicitly include corners like the RenderTransformMesh does, so while it gives expected results under certain ratios of height and width to meshCellSize, it often does not match the bounds of the RenderTransformMesh.
This commit uses similar logic to the RenderTransformMesh class to uniformly sample between corners during sloppy bounds derivation.
The current sloppy bounding box is inconsistent for x and y (x<=width, y<height) and does not explicitly include corners like the RenderTransformMesh does, so while it gives expected results under certain ratios of height and width to meshCellSize, it often does not match the bounds of the RenderTransformMesh.
This commit uses similar logic to the RenderTransformMesh class to uniformly sample between corners during sloppy bounds derivation.