Hi! Here my solution, nice challenge!
Fun Challenge! Thanks for the idea @CharlieS
As for the tiny triangles, they would be collinear if drawn on a flat plane, as CharlieS mentioned, but since they're mapped to a sphere, they end up having a small area. Here's a plot of the 24 thin ones
Here's one of vertical ones zoomed in
The length(height) of this triangle is 19.3 km, but it's width is only .008 KM.
I was thinking the same thing in that non-collinear would mean the opposite of Collinear, which is usually defined as all points on the same line. Furthermore, when dealing with modeling, co-llinear is generally referred to as non-desired trait in a model as you likely have two variables that that are essentially the same, just changed by a scalar. A good example would be weight and volume. If you have the volume of water in Gallons in your model and you have the Weight in pounds, you will have two variables that are Collinear (Water weighs 8.5 lbs per gallon assuming the temperature stays the same). With geometry, I understood the definition of a Triangle to mean that it was Non-Collinear in nature...
-This lead me to originally think that I would have to find all combination possibilities and then eliminate those that were CO-Linnear.
Anyway, semantics aside, I had a hard time with this one, as i had to peek at others examples, as I got stuck with way more than 516 Triangles.
Excellent Workflow that you have there!
-I had a few questions regarding the Sort and Subsequent Multi-Row Formula.
-Why is the Sort Required for the Workflow to run? What changes as I got stuck with something like 1800 Triangles...
-Multi-Row Formula with OrderID: I really like the idea of the Internal Order piece as I think that is a great method of having a Grouped Id for all kinds of applications.