I've heard that Delta DOR can be used to get spacecraft direction, but requires more antenna resources than ranging or doppler.
I don't know about that. It does help to get rid of systematic and modeling errors in the range measurements, but even over a fairly long arc, I think the thrust effects would only be observable as an angular offset: the range wouldn't be noticeably different until you'd had a lot of thrust events, spread out over enough time for it to show up in the ranging measurement.
Which is to say, I think it would take longer data spans than these guys were probably using. Also, it sounds like the NAV guy wasn't knowledgable enough about the SV to understand what was even going on to cause the effects.
As the report says, though, as early as April the data did show that there was a systematic modeling error, but the NAV guy didn't adequately express it to the rest of the ops team. (The vehicle was lost on 23 September.) By that point, they had corrected the small forces software, knew there was an error, the flight team had plenty of time to respond ... and didn't.