Cross-check with Yoshida_2016 paper (Fig.5):

Using a script here I plot the image
https://gist.github.com/Sheshuk/3a11f9b033fb8a3029b53fada9aa4f9a

From the figure, it looks like the NU_E and NU_X rates are almost identical (both solid lines overlap); while the NU_E_BAR and NU_X_BAR are both almost identical to each other (both dashed lines overlap), and a factor of ~4 lower than the solid lines. However, physically, this doesn’t seem right at all: The main emission mechanism is pair production; so NU_E should be very similar to NU_E_BAR, NU_X should be very similar to NU_X_BAR, and there’s no a priori reason why NU_E and NU_X should be similar.

Therefore, I suspect that NU_E_BAR and NU_X have been mixed up here. Looking at the README, the input files use the order NU_E, NU_E_BAR, NU_X, NU_X_BAR; while the Flavor enum in snewpy.neutrino uses the order NU_E, NU_X, NU_E_BAR, NU_X_BAR. Maybe that’s the origin of the problem?

Oh, you're right, I mixed up the flavors! Thanks for noticing this! Fixed now, here is the image from the same script: