One hard lesson I take away from this project is that spurs are the inevitable consequence of ANY multiple-conversion scheme. You can attenuate them as much as you want, but they will always be present at some level. That being the case, the best way to get rid of them is to avoid multiple-conversions altogether, especially in an integrated receiver, and especially on a single-dish instrument. That wasn't an option on the prototype array since we wanted to multiplex the IFs, and a single downconversion to IF less than ~6 GHz wouldn't provide adequate image rejection. But for the 60 element array, the IDM2's (or another single-downconversion approach) could be used with an updated IF Transmission System and the spurs would no longer be an issue.
There is way, however, by tuning both LO1 and LO2 together, to avoid seeing any spurs of any order in the IF passband: