Sorry for the long delay, it's taken me some time to get back to this.
At this point I'm really stumped. Neither of the above options (either altering the k.Series library code like I proposed in my first post OR changing the address using the code that Jason posted) work for me. Using the corrected version of my code from the first post one sensor always reads at around -290 and the other one reads correctly. This remains true even if I switch which sensor is connected to which pins, so I know it's not the sensor or pin configuration.
In Jason's code the change address part runs fine with one sensor attached. Then, with the chance address part removed and both sensors attached it doesn't get past the sendRequest(readCO2_a); line. I assume that this is because it can't contact sensor A so it get stuck on the loop in the sendRequest section. All I changed in this code was the first line of that section to packet as per dc42's suggestion.
The fact that Jason's code didn't work made me suspicious that the address change didn't really do it's job when I ran it. The only way to test this that I could think of was to separately try both sensors using the example code (written for a single sensor at the default address). Since that code did read the CO2 values from each sensor (separately of course) I take that to mean that the changeAddress part did not work.
I'm at a loss, at this point, for what to do next. The changeAddress code is way beyond me so I wouldn't know where to start diagnosing that. The only other option is to use the GasLab software like Jason suggested, but that seems a bit daunting and I'm not even sure that that will solve the problem (what if it's not the addresses that are the issue?)
So, any suggestions?