Can you put the O2's in header primaries?
 

Would this make any sense if you had an actual full length header(34" primaries or so). You'd only be monitoring that one cylinder, yes, but you could get a better reading than having the O2 in the collector couldn't you? Just an idea/question...

Notice also that in the factory manifold the O2 is after the gases are collected which would yield a better average of the true fuel mixture being burned (rather than just one cylinder)...which is why there is a Bank 1 sensor and a Bank 2 sensor.

Well now you can't compare it to stock manifolds mang. I'm talking about full length headers. I realize that if the sensor is in the collector it can get the best average reading of all 4 cylinders. But if the O2 is, let's say, approx. 34" away from the exhaust port, it's getting a vague reading. Now if you put the O2 in one of the primaries lets say 3" from the exhaust port you are getting a better reading, but just for one cylinder. Just an idea...

Its the exact same idea only the collector is much closer to the motor on the stock manifolds.

In fact your idea is closer to the stock manifolds or a mid length header as far as position of the sensor....but as said before the reading from one sensor is NOT accurate enough to assume every other cylinder would be the exact same.

I think a long-tube header, with a balance or
sampling tube that ran across all 4 pipes with
an O2 bung in the center, could place the O2s
"close enough" to stock location and give a
balanced average reading. With enough effort
at arguing with the CARB people you might even
get them to buy off on it being "same location
as stock" although that still leaves the cats as
a "gotcha".

However that's like another dozen+ circular pipe
welds to make and not have break, time is money,
give-a-shit factor is low (at least until the first
guy to get the scheme past CARB starts stealing
the California market).