Some years ago, the RAF had a not dissimilar problem with a handful of Shorts Tucanos which would, if stalled with full flaps, roll inverted in about a second. On the whole, if you were nibbling the stall on finals, that was arguably an undesirable characteristic.
So, we spent a fair bit of flight test time and money trying to solve the problem - and failed dismally.
Then a squadron Engineering Officer and Unit Test Pilot somewhere got together decided to swap front and rear fuselages around a bit and fly them. Mysteriously, the characteristic disappeared - so we stopped flight testing and got on with something else. My best guess is that the basic problem was cumulative manufacturing tolerance (given that you could get about 10" variation in the length of a Tucano, and about 2" variation in headroom between aircraft, this wasn't hard to believe).
I've seen a few other vairations upon a theme - although never quite that bad again. This leads me to suspect, if all the controls are in tolerance is that your actual airframe is bent. Not necessarily badly, and probably within tolerance, but just occasionally you can get an airframe where all of the manufacturing tolerances can add up to give you something quite undesirable.
That said, this is pure guesswork on the basis of very little evidence. If you can post some detail of what specifically the pilot didn't like about it, and at what conditions, it would be helpful.
G