It seems to me that in a dual extruder setup, that each extruder should be separately calibrated for e steps/mm, and that the firmware should allow multiple values in its configuration.
The main reason that e steps cannot be accurately calculated is because it all depends on the geometry of the hobbed bolt. All other factors - gear ratios, etc - can be solved algebraically, but the variability of the hobbed bolt (or whatever gripping technique you use) introduces uncertainty such that the iterative calibration procedure we all know and love is necessary. It follows from this that every extruder is unique, and that uniqueness is captured in the e steps/mm value. Thus if you have two (or more) extruders, you should have two (or more) difference e steps values.
I have a dual extruder printer where I see this issue. Mine is a very small difference because the extruders and gears are identical (except for being mirrored). My difference is small enough that splitting the difference is probably going to work fine, but I can see that this could potentially be an issue for others.
Am I missing something here?