I've seen the same in my model, and when I played with your model I've seen the same thing even with the errors in said model. The more currents flowing on said section the stronger the horizontal component seems to be. If there is no matcher, everything that is horizontal is in balance so there is little to no horizontal radiation.
I haven't looked at it closely yet, but I saw the same effect, the trombone matching section increased gain slightly in my models.
This is something I haven't noticed, as of yet, but then I keep finding myself doing other things with this and other models... It is still on my list of things to do.
A direct comparison between antennas including matching system is what I wanted to find out, and part of why I started making the i-10k model. I figured that with what I know, of all of them it would be the easiest for me to build and tune...
While there is a difference between modeling with and without a matching section, I have yet to see a big difference, unless I intentionally added extra loss using the matching feature in 4Nec2. From what I understand, that feature uses ideal components, which will show less loss than what would normally happen in reality. I actually tried to make a model of a half wavelength antenna that had the bandwidth of an a99, I never succeeded, and even with half of the bandwidth gain was down by well over 3 dB. I think that was a bit extreme based on people's actual comparison reports. Further, said matching devices also seem to affect the bandwidth you get from them. I can't just plug one in and assume its results are close unless I know a matching device was made based on said data. Unfortunately it isn't really useful for most antennas, it seems to be more meant for making such tuning systems for antennas than antenna comparison use, which is more or less why I took up modeling to begin with.
Sure.
The DB