Rough fix for issue96

[stefannikolei]

This is just a rough fix for the last test failure of the js generic tests where we have a mismatch in the count of points.

This is also done in the rust version. mod.rs L:116

Do you see a better way how to handle this? Perhaps change the return value of ConnectEdge to Contour[] and then do the ordering?

When this is done only 2 tests of the js test suite are failing. And those faile because of probably some rounding problems

[JimBobSquarePants]

@stefannikolei I'll have a look at this asap. I've been preoccupied looking at some gnarly ImageSharp issues.

As a side, I've been porting more of the Rust version code into a local branch. There's a bit of difference in that codebase around segment comparing plus some intersection calculation changes. It doesn't fix things yet but it may make a good starting point to getting everything working so will push the branch tomorrow.

[stefannikolei]

@JimBobSquarePants I looked further into the remaining failing tests. I get a difference in the connectEdge logic. Our implementation for example return 8 as the nextPos. But the Rust implementation returns 9.

The rust implementation is using an precalculated iteration_map. I think our focus should be on this one.

I will wait for your remaining rust ports and then look further into it.