I tried using a float for a simple example and found that the rate variable uses rate[rate>250] pattern now:
I would like to change this to np.clip(rate, -np.inf, 250.0), so that it works with scalars, but I thought I'd see if anyone sees a risk in this before making a PR.