Yeah, I would think you need to have the transmission in neutral. Seems like you'd be getting "engine braking" otherwise - unless you KNOW your torque converter "un-locks" when coasting. ???
In my mind, it's just too many variables to really get an accurate estimate of Cd from coast-down testing (although I've got to admit, I haven't tried using the formula - spreadsheet - so I don't really know what I'm talking about - LOL)
I have had some luck with coastdown testing in older cars I couldn't hook a ScanGauge to. In that case, I also used a GPS for speed, and went to a deserted road where I could start the test at over 100 mph (usually 105), and then coast it all the way down to 55 or 60. I'd do multiple tests, both ways of course, and get an average. Then repeat with whatever aerodynamic change I was testing. Kind of tedious, and quite a bit of variation from test to test, but hopefully, the averages "averaged-out" enough to get useful data. Usually, there was a clear "trend" in the final numbers that I was happy with (confidence I had "useful results")
But I never attempted to get any actual Cd numbers from it - just to see if "this is better than that". If it was, I'd use it. And it was in connection with a race car we were running at Bonneville Salt Flats for top speed. I had another car of almost the same body (2 years different) that I'd take out and test the aero mods on by coast-down testing. I eventually got up to over 182 MPH on the salt before I went broke! LOL
(pictures of the "PLANN-R Research Car" attached)