Some years ago, I mounted a Starduster above my 4 element horizontal, which I felt worked very well for me. I was surprised that I did not have to re-tune my beam after mounting the SD'r so close.
Due to the design of the Starduster, it needs a mast that is at least 10' feet long so the radials can be clear at the bottom. These radials hang down about 8' feet below the hub were the mast is attached. This allows for a foot or two of clearance. The way I mounted the antenna, the 10' mast also lost a foot of length in mounting, so the SD'r radials ended up very close to the elements on the beam. Initally I was concerned that this setup would affect the beam's match, performance, and maybe even the bandwidth, but I noticed none in operations with both and the SWR bandwidth remained the same. This was before analyzers were reasonably available, so I was not able to compare at that level.
When I modeled the setup I posted above based on BM's question, I was again surprised to see there was so little difference. I admit I did not attempt to make my models anywhere near to specs...by any means. I just put the two antennas together, and like I said earlier I used the beam model provided with my Eznec 5 software listed as: "11m5elyagi.ez"
Henry, you no doubt have more experience with beams, and for sure you have more experience with modeling, so, unless you are talking about the minor issues seen in my model posted above, could you post here the wires view, PM, or Email me a model that supports your claim posted above, where such a setup would have issues?
The model I posted earlier, pretty well confirms what I think I saw in my real world experience, and I would like to see what it is that you are considering in your claim that such a setup with resonant radials would really mess things up.