Great, I saw your images.
Yes, I noticed that the far-field plots had <1dB difference; I don't think those scenarios are going to be detectable outside of laboratory conditions & instruments.
About your graphs though, the modeling software I've used, EZ-NEC & MMANA-GAL, don't consider source impedance or reactance in their gain calculations, so if you are modeling for a certain radiation pattern of an antenna at a given frequency, to actually get that calculated gain, you need to match the feed-point impedance of the antenna to the rest of your system, are you doing that?
(For people following along) In other words, if you want the 2.15 dBi theoretical gain of a 1/2λ dipole over an isotropic radiator, for a 50Ω system, in practice, you'll only get 2.15 dBi gain if your antenna feed-point impedance is also 50Ω; If you have an input impedance of 73Ω (which is what a dipole has in "free-space"), that would create a 1.46:1 SWR, and your gain would not be 2.15.
So I'm curious if that is taken into consideration in your graphs of Gain & SWR?
