This bug is unusual, to say the least...the power comes from the crankshaft to the mainshaft (those are the gears in the upper half of the case) via the primary chain, then that shaft puts the rotation (power) into the clutch outer. The clutch plates couple that rotation into the mainshaft, where the splined-on gears get it to drive the gears on the countershaft. There are several "idler" gears, not splined to the mainshaft, which couple to the splined ones when they are slide sideways into the splined ones, by the forks. This is how the power is transmitted to the countershaft.
The countershaft has a large gear at the very end, near the center of the lower case, which couples directly to the final drive gear. The combination of gears that pick the rotation from the mainshaft, then couple that rotation to the countershaft, determines how fast this final gear spins, and it is always coupled to the final drive.
In other words, it is a "constantly meshed" gearbox, where power transmits from the dogs on the sides of the gears into the holes on their adjacent gears, to the shafts.
So...about the only thing I can think of that would let it freewheel would be that the middle fork is not engaged on the mainshaft, and the gaers on that shaft happen to be just barely in between two gears (2 & 3, probably). I seem to remember having done this once on a 350 twin (similar gearbox design) years (and years and years) ago. After that little incident I always run through all the gears upon reassembly, before I tighten the crankcase screws down on the sealant.
Sorry, I can't think of anything else at the moment!