Interesting - as I'd understood it, it was always dependent on reliable signals from the O2 sensor - which doesn't have anything to do with ECT readings, since the HEGO isn't anywhere near coolant.
Also, you say the downstream sensor. Stop me if I'm off base - but the downstream doesn't have anything to do with fuel trim - it was added in OBD-II to monitor catalyst efficiency, no? Since the catalyst isn't heated, they could use ECT warmup times to "guess" when the catalyst should light off, and then it could accept signals from the downstream HEGO - but fuel trimming and monitoring is done with the upstream HEGO in the exhaust manifold. Correct?
I suppose it's possible that timing of when to monitor the HEGO could be done by using ECT readings, but that would seem to be unreliable to me (if it always lights off at the same reading on the gage, you'll have varying light-off times, depending on how long the engine had been sitting and ambient temperatures. If you use "so many degrees above ambient," you still have the issue with thermistors having a relatively wide tolerance range, as I've usually seen it. So, that doesn't make much sense to me - it would seem to be better to use a heated sensor tip and have a timer decide when to go into closed loop. EGO sensors used to light off after about four minutes, while HEGO sensors seem to be able to light off in thirty seconds or so, as I recall.)
Sounds like this is something I may have to dig into some more...