RIP. I respect the author for the effort but source CAD kernels are a tough nut to crack; the existing solvers are pretty old at this point and there's a floor on the effort to get a kernel that's roughly on par with opencascade without the architectural limitations. We're going to be waiting on an entity like CERN or other organized body to put in the effort to meet that floor.