Thanks for the suggestion.
What Steve Teeple did was to take advantage of the fact that there were no “Position Key Frames” in the leg-part of the Motion Clip. Hence the position changes were not overwritten, as there was no such information to beginn with. (Doesn’t work there with rotation, as that information is in the Motion Clips.) A clever “hack” by itself, but anything key-framed should be done after the motion clips have found their position, otherwise one has to pay a great deal of attention to keep both (clip and keys) in sync.
Kai did an introduction to the Animation-Layers when they were released:
This suggestion is a different way than discussed in the series you mentioned. First the Motion Clip is converted to Keyframes, while the all layers are switched off after this. Then a relative Animation Layer is used to get the fixes done as needed. Then turn anything into a new clip, while deleting the object keyframes as well. This can be now used freely. Save the clip to a library. (Don’t forget to switch the Animation Layer off after baking! … if will add otherwise the values “again” to the rig.)
My suggestion would be, to always try to stay in the Motion System (vs keyframes from the the Object), and turn every time the results of the fixes into Motion Clips again, to stay flexible and open with the idea of the clips and their opens to be mixed and moved.
My best wishes.