Yes. And a salient point.
I hadn't considered that the half speed condition was the result of a series fan configuration.
What would happen (ideally) would be the lead (positive side) fan's shunt generates 6.01v, more than enough to keep the first comparator off. The downstream fan shunt would generate 12mv, which is still high enough to keep that comparator in the "off" state (anything >8mv).
Now, with the fans in series, if either or both fan motors fail, the entire circuit will cease to draw current. As a result, both comparators will go high, lighting up both LEDs, a pretty obvious failure indication. It would not, of course, indicate which fan failed.
If need be, the comparator circuits could be modified for a slightly lower trigger voltage by reducing R3.
Noise tests in SIM (12V [24v P-P] sine, triangle and square waves, sweeping 1 to 1kHz over 1s, injected at sensor input to comparator input) had no significant discernable effect on comparator function. Only the negative going transitions had marginal effect and were generally too low to trigger an LED response. I think that this largely due to the shunt(s) essentially shorting the noise to ground. This may not reflect real-world conditions.