Well, Mythbusters is well-known for very sloppy scientific work so I'm extremely skeptical of their result. That being said....
Changing the surface texture characteristics of an object to create a local turbulent boundary layer is a well-known way of reducing aero drag in some situations. The idea is that you want turbulence to be confined to a thin layer rather than having drag-intensive large vortex shedding, so you introduce local surface roughness to encourage the early formation of a thin turbulent boundary layer. But whether this technique works is highly dependent on a number of things - it is not a universal "fix". Boeing spends millions trying to reduce the drag of their airframes by a few percent with each new generation of airliners - if they could get a 11% reduction simply by dimpling the exterior, I would think they would have done this long, long ago. And in any case, you weren't talking about dimpling the entire exterior surface of your hardbags - you were talking about adding aerodynamic tweaks to the edges.
I'd encourage you to try some things out and see what you think. One way to assess your results would be to top-end your bike on the same deserted section of road before and after the tweaks. The extra speed would amplify the difference. But you'd have to gather a fairly large amount of data to damp out the confounding variables (e.g., wind).
If you really wanted to do this right, you'd use an instrumented wind tunnel (either physical or computationally) to do your design work. But, as I said earlier, even if you did discover a tweak that made a significant difference in drag characteristics, the fact the bags already sit in extremely "dirty" air behind the front bodywork of the bike and behind the rider's knees would likely reduce the benefit to the point where you couldn't measure it on the bike in typical conditions.
Making the overall shape of the bags radically different (e.g., giving them a long "tail" to keep the boundary layer attached for a longer period) is, IMHO, the only thing that would likely result in a measurable difference. But that's just my fluid mechanics intuition (and 2K+ hour pilot intuition) talking so take it FWIIW.
- Mark