I can't believe I read the whole thing.
My take: Wow. This is an amazingly thorough, well thought out proposal for a completely new product. If the original IC2 power system had been implemented this way, that'd be awesome. But it wasn't... so, what to do?
One does not simply retrofit the innards of something this complex. Even if I had the code, I wouldn't want to touch that particular beastie unless it was a crashing bug. As it is, IC2's power grid is amazingly efficient. Even with some 50 generators talking to a few dozen storage devices powering 0-20 machines at a time in my server's central hub... there is no performance degradation. The game continues to run well.
Yes, the system isn't realistic. Yes, it takes some learning. Yes, using an EU reader on the output of a water mill is humorous. But as far as the players are concerned, the learning curve is not terribly steep.
I think the efficiency of cabling should be readdressed... and that's all I think we could possibly expect as part of the core module. And really, the only reason cabling efficiency is even an issue is the problem of long-distance transport of power - to which problem there are a variety of lossless options (though, admittedly, none are terribly elegant as the standard superconducting copper/transformer chain example illustrates). The ~20% loss observed when using copper to transmit LV for 20-30 meters only significantly impacts players at the early stages of the tech tree. An ULV transformer would be more than enough to solve that problem for them.
The only realistic way to introduce the full scope of changes you are proposing is to write them from scratch. Be that in an alternate power grid mod or a fork of the project... that's what it would take - and that's bad juju there. Personally, I am willing to run rail between the nuke plant and the central distribution hub and let energy carts do the trick - it's a more fun solution than just knowing I can string a lot of high throughput cabling all over the place and forget about it.