AAA poles
I noticed that one transfer function that I fit with AAA did not have pairs of conjugate poles. I attached a simple example. Increasing the data that I fit (up to 1000) points did not fix this problem, though it might change the number of poles. (I've yet to encounter a scenario where the zeros don't come in conjugate pairs.) I tried messing around with s_tol and res_tol with no effect.
I can understand why the number of poles might change, but I'm not sure why they wouldn't come in conjugate pairs since I thought the analytic properties of transfer functions are enforced with this fit method.