Such a fun topic. And then there are some things tunnels can't do, or can only do in the most limited form, like analyse post-stall behavior. You can't very well have your model tumbling free in your flow, it might crash and make a mess. (Well, there have been attempts, some of them successful, but also quite limited in their parameters.)

So you have to resort to a free flight model, scaled down to simulate the parameters required. There's a (fascinating, for those so inclined) treatise on it here[0].

[0] https://ntrs.nasa.gov/api/citations/20110012492/downloads/20...