Very short recap: I believe that a single act of copying goes wrong when the cumulative weight of distracting forces is bigger than that of the focussing forces. This means that often there may not be one single factor (or 'explanation') that accounts for the creation of a textual variant, but rather a number of factors working in tandem to tip the balance.
Here is another one of those scribal tendencies, misreading αυτου for τουθυ or τουχυ and vice verse. Visually there are two syllables involved (not of course in the full nomen sacrum του θεου, or του χριστου), and such metathesis when reading is not uncommon at all.
A few examples (all [sub]singular readings in order to demonstrate the phenomenon):
Alexandrinus in Eph 3:2, της χαριτος αυτου (for του θυ)
P46 in Eph 3:7, της δυναμεως του θυ (for αυτου)
Minuscule 33 in Eph 3:8, πλουτος αυτου (for του χυ)
In each of these examples there is the additional factor of harmonisation to the immediate context, but the combination of the two has of course more force.
(Other examples, 69 (and others?) Rev 14:10 θυμου αυτου (for του θυ); 1854 Rev 20:4 λογον αυτου (for λογον του θυ).
What is perhaps most instructive is that this phenomenon underlines that even though Greek is written without word breaks and that there is, therefore, a large phonetic component needed when reading a text, there is still a visual component too that also works on the level of 'syllables' or perceived syllables.