when using hulls to calculate Delaunay then you calculate everything in d+1 dimensions, which for 2D has a 4X slower predicate. Also predicates are mostly precalculated (already implemented in the updated gist above), so their repeated calculation in the "swap" method should be fast.
Anyway, let me finish a robust 2D Delaunay implementation in Julia and do some benchmarks against [CGAL](https://www.cgal.org/), [QHull](http://www.qhull.org/) and [Triangle](http://www.cs.cmu.edu/~quake/triangle.html) Should be interesting :)