Triangulate bug: Two separate polygons about manifold

[kintel]

I opened a PR with the failure case. The precision setting isn't needed, I just inherited the argument from surrounding tests in my first repro.

I experience lots of errors like this in OpenSCAD (first attempt at using the triangulator). This is the most trivial one I could find, but it's probably good to start off fixing this before potentially registering a bunch of duplicates.

[elalish]

When we run our tests, we run them with params.intermediateChecks = true, which asserts the manifoldness of each triangulation, and with params.processOverlaps = false, which checks that every triangle is CCW. Those should catch this well before the number of triangles check - would you mind adding this test as a PR; then we should see that behavior on the CI.

This seems like a pretty basic error - I'm surprised our existing corpus doesn't catch it.

[elalish]

Not that it should matter, but that's a surprisingly wide precision you've specified - is that necessary to repro?

[elalish]

Closing as expected behavior since this is an invalid polygon.

[kintel]

Thanks for taking a look - I'll fix my upstream issue and run my larger tests :)