I'm no expert and can only go on what I've read. This book
suggests a method involving glide tests. The idea is that in order to construct a drag curve you need various things you can measure (mass, wing area, aspect ratio) and two "unknowns" which are the parasite drag coeficient, Cd0, and the inducued drag efficiency factor. If you knew these you could construct the traditional quadratic drag curve and hence predict the best glide angle and the speed at which this occurs.
If you do some glide tests over a range of speeds, you should be able to determine the best glide speed and corresponding best glide angle by trial and error. You can then work backwords to determine Cd0 and the efficiency factor.
Of course this still leaves the problem of shutting down the engine(s) and accounting for any residual prop drag. It also assumes that the traditional drag curve is a good model.