Vector

In this scenario your position is already near the equatorial plane. In this case there will be a cheap inclination burn that will level out your inclination, yes. But if you are approaching Eve "from above" then the pink normal burns do not give you a cheap equatorial orbit. They just let you pick which polar orbit you want. Off to take pictures...

I will post some screen shots later tonight showing the regular inclination-adjustment (expensive) method and also the double-encounter method, which I tried out recently. I've uninstalled MechJeb because I was encountering some bugs (orbits drifting under no thrust and even when game is paused) and it turned out the bug went away when I removed MechJeb!

Thinking about this a bit, I realized it is also possible to use the planet itself to level out your inclination, but it is even more challenging than using a moon like Ike. The idea is that you would do a non-capturing fly-by of Eve, entering at the inclined angle and leaving in the orbital plane of Eve. This is the easier part. The harder part is you have to adjust your exit such that your orbital period is very close to the orbital period of Eve, so that you can get a second encounter after one trip around Kerbol. On this second encounter you will be in Eve's orbital plane already. This is going to be very challenging, and it takes another trip around Kerbol, but it will be very efficient if you can pull it off.

I too have noticed large inclinations when I reach Eve, and it is not the SOI transition that does it. The reason is the inclination of your approach does not match the inclination of Eve's orbit. I see you are going pretty much straight from Kerbin to Eve and aiming for the AN or DN without adjusting inclination, (which is in my opinion the best way to do it). But this approach leaves you with a large difference in velocity between yourself and your target in the "vertical" direction (North/South, or normal as they call it). This large delta-V in the normal direction means you can never have a low inclination orbit when you arrive. Inclination changes are costly, but if you were to do an experiment where you first match inclination with Eve's orbit, then I think you will find that when you reach Eve you will not have much normal velocity relative to Eve and your orbit around Eve will be nearly parallel to Eve's plane around Kerbol, or can be adjusted with small burns to be so. The inclination with the target planet will have to be very closely matched, because for planets even small differences in inclination correspond to really large delta-V at the ascending and descending nodes. If Gillie weren't so tiny you could use it to cancel some inclination by swinging around "above" or "below" and leaving Gillie in the plane of Gillie's orbit. Gillie's orbital plane is closer to equatorial and would be an improvement if it were possible, but Gillie is not nearly massive enough for this to actually work. I mention it because this same idea does work on other planets.

I'm taking a crack at this problem. So far I've got the forward modeling working, meaning simulation is pretty close to experimental results. For now I've left out Coriolis force and the rotation of Kerbin but everything else is in place. The representation of the problem is a huge factor in the optimization, more important than (or really an integral part of) the optimization algorithm. I've got two or three failures so far where my representation/optimization failed to converge or where the discrete approximation made too much of a difference. All this is in a brand new C# program so hopefully there will be avenues to construct a new plugin or integrate with MechJeb.