T-numbers are not the be-all, end-all here. They are most useful in the deep tropics where the observational network is sparse, at best, and where satellite is the only tool from which a forecaster has to draw. In the case of soon-to-be TD3, we have a relatively dense surface observational network and a few radars to observe the system, and they are telling us that we do indeed have a tropical depression, despite the low t-numbers. Direct observations ALWAYS trump t-numbers; not classifying TD3 due to poor t-number would be like a chaser following a tornado on the ground but the NWS not issuing a warning cause they can't see the circulation on radar.