Incorrect calculations of a centroid position

[daniil-loban]

I noticed that the centroid for the Bounding Box node is calculated incorrectly when there are multiple shapes passed via Flatten Path.

[0HyperCube]

The area_centroid_and_area functionality does indeed appear to be broken when applied to a rectangle with linear segments (as seen in the test below):

let rect = Subpath::<PointId>::new_rect(DVec2::new(100., 100.), DVec2::new(300., 200.));
let (centre, area) = rect.area_centroid_and_area(Some(1e-3), Some(1e-3)).unwrap();

The current implementation returns centre as (800.0, -50.0) which doesn't really make any sense to me. I'd expect (200., 150.).

I kind of fixed the implementation to make it return (200., 150.) by changing fn pathseg_to_parametric_polynomial to return a cubic polynomial for a line segment (the equivalent of cubic bézier with the handles on top of the anchors — see attached diff).

I presume there is an edge case when the derivative is constant?

However I don't know what I'm talking about, maybe @elbertronnie would be able to provide more insight into the algorithm?

change_line_polynomial.patch