Dual and quad cells return all the iron used in crafting them when you centrifuge the depleted parts... so only quad cells should have a running cost in IC2 experimental, and only one copper plate per quad cell (down from 40 in classic IC2 or 8 in GregTech). A much-needed improvement!
I really like your upgradable reactor path! The four single-cell ones can all be converted from one to the next larger, without having any components left over. They're very diamond-heavy, but you don't have to spend them all at once - the cost is mitigated by allowing you a lot of time in between upgrades to procure more. Decently efficient, too; I haven't really found any good designs that go much beyond base 4, due to troubles spreading the heat fast and far enough (kind of similar to the "not enough sides to a block" issue you often have ingame). So up to 3.6 isn't half bad.
I suppose a reactor using core transfer as described in the opening post could achieve higher efficiency but would have usability drawbacks because it couldn't maintain its heat level while switched off or out of fuel.
Oops, completely forgot about the centrifuge giving the iron back!
So after some more testing, reactor plating has no effect on boosting EU/t any more, it seems to be all % of max heat now. I have now upgraded to the last reactor I listed, running on the dangerous side of 9995/10000 heat, outputting a nice 900EU/t. External cooling or hull transfer reactors could probably easily output more, but this is pretty maintenance-free, although diamond heavy (but by the time you've spent ~40 diamonds on an MFSU I doubt another 24 is going to be a huge problem).