I think what he means is the expected $ you will have at the end of your turn when starting from a random 5-card hand at the beginning of a shuffle for a given deck.
$Power is this number divided by 5, which I guess corresponds to the expected pay-off per card in hand, which is a flawed way of looking at things because it presupposes a linear scaling that is absent in reality.
I believe this term has been called "money density" around here before. It's been talked about but as SCSN hints at, doesn't apply to most games of Dominion, since it really only applies to Big Money strategies which aren't dominant on most boards. Your formula could potentially be a more accurate calculation of money density than what's been done here before.
Results you can get for this can be useful, you just have to be extremely careful with how you apply them. Simulation and bots that play Big Money seem to be the main applications for this. I'm not an expert at those so I won't say more.
If you want to calculate this for more complicated decks you should just write a Monte-Carlo simulation, because the reason you're having big problems with it isn't because it's hard but because it's extremely cumbersome, like manually counting the number of letters in the Bible.
In any case I think this approach is an exercise in futility because Dominion isn't at all about maximizing your expected coin production, and a Bot utilizing such a mindset should be classified under Artificial Stupidity.
This is completely unnecessary. Just because you're better at Dominion than most people doesn't give you the right to be dismissive like this. Grow up.
I used to think this community was above stuff like this, I used to think we welcomed people and were only critical of their ideas in a constructive way. It seems we're having trouble with that recently.
