OK, lets start all over again...and lets start with the offense since those numbers are less guesswork.
Fontenot vs. RHP
.298|.379|.473 - 6.92 RC/27
Keppinger vs. LHP
.351|.403|.515 - 8.65 RC/27
Platoon (figuring Fontenot gets 2/3 of the ABs) = 183 RC in 660 ABs.
.290|.369|.422 - 4.76 RC/27 = 116 RC in 660 ABs
Difference = +67 for the Fontenot/Kepp platoon.
BTW - I used this site http://www.tangotiger.net/markov.html to calculate RC/27 and I used the Bill James method.
So the Fontenot/Keppinger platoon would have to give up 67 more runs in the field than the average SS to break even.
Now for the defense...
The 32 runs/game is not generous. If anything, it's high. Like I said, total chances overstates the run value because it counts assisted plays twice. A groundball to the SS = 1 assist for the SS and 1 PO for the 1B, which = 2 TC and only 1 real out. INOW - suppose there are 7 Ks in a game; that leaves 20 outs in the field. However, that same game might have 30 TCs because 10 of those a assisted plays. So that method is flawed.
What we should do is just use assists alone for a middle-infielder. The reasoning is that a great infielder probably won't make many more POs than an average one. Anybody can tag second base or catch an infield fly. However, he will make a lot more assists. Then subtract errors.
The problem is, again, I don't know the averages for a SS or a 2B. FWIW, Fontenot averages 2.71 assists/27 @ 2B and Keppinger 2.62 assists/27 @ SS. The average assists for a SS and a 2B tend to be about the same league-wide. Assuming that 4th out = 1/2 run (your assumption), the average SS would have to average roughly 3.5 assists/game to make up the difference in runs that Font/Kepp give you at the plate. The league leader in assists (Orlando Cabrera) only averaged 3.05.
So, even using the defensive league leader, the Kepp/Font platoon still comes out on top by 31 runs. Even if you assume Fontenot can't play SS as well as 2B, that's more than made up for by the fact that I'm comparing him against the league leading SS, as opposed to the average.