One of the thoughts that came to my mind is that if the flow solver fails with the deformed grid using my integrated-FFD method, it isn't worth a damn. So I fed the deformed grid results from the winglet case into the group's parallel Newton-Krylov-Schur flow solver, and it worked. It was really a sense of relief that yes, things are still okay. Here's the surface pressure coefficient CP plots from both the flow solutions with the original grid and with the deformed grid. The case as a M=0.74, with an angle of attack of α=2.0°.
The lift coefficients CL and drag coefficient CD are of course different for both geometries, but in case that we are worried that the values are inaccurate, we can always refine the grid and re-compute them again.