Im wracking my brains here but this scenario sounds like a very early HW3 code option that thankfully, seemed to have been dialled out. Hopefully they will have learned from this and can do the same this time around.
I personally never suffered with the issue, but my wife did and it was a long time before she returned to using TACC.
Based upon some empirical evidence from her and others, is that when overtaking, the car would speed up to start the overtake (as it is designed to do), but because of the follow distance to the car in front (ie the one being overtaken), its as if the car seemed to think it was going to drive into the car being overtaken then suddenly change speed (some described it as a phantom brake). The solution at the time seemed to be to give a bit more room to the car in front when overtaking.
However in this scenario, the slow down was part of the overtake, where as with yours, the overtake may have completed? If so, doesn't sound like the same issue, but may be worth having a think about how close you may have been to the same car when you started your overtake.