Note (update): These initial measurements were measured with poor-quality probes, which was the cause of their peculiarity (spoiler alert). See the update at the bottom for more information.

Before installing the transformer T1, the impedance that the amplifier at the RF output of the mixer was high enough that and RF signal was seen at the output of the mixer:

However, the LO signal is too low to produce a measurable signal at the RF output of the mixer now that the transformer has been installed.

This is because the DS90LV028ATM is only providing about 580mV peak-to-peak, as opposed to the ~1-1.5 volts I had expected.

Luckily, the board has provisions such that this signal can be boosted by a level shifter:

Changing JP5, JP8, JP9 should allow the LVDS receiver's output to be boosted to a 5v logic signal. However, the LVDS receiver's outputs are not compatible with 3v logic in this situation in terms of voltage, being centered around 1.64, which should appear as a high. It is likely that one of the outputs is being dragged down such that neither works properly.

UPDATE: As it turns out, the cause of this symptyom was poor measurement. When I switched to 300MHz Rohde & Schwarz probes (as opposed to unknown probes), the signals behaved as expected.