There are three 480V primaries, one per core, each one has a 400V and a 440V tap, a few turns before the end.
I easily managed to free all coil ends and wire them as I proposed above, the output-coils-series picture.
230V (measured) mains in the first core primary 480V coil.
Second (closest) primary 480V coil shows 193V.
Third (farthest) primary 480V coil shows 35V ...magnetic flux clearly do not split in equal halves.
Second and third in series shows 228V.
I am already good like that but my feeling is that Tiziano's suggested scheme is better, with mains in outer coils in series and output from the center one, where each external core can have its best influence.
Last thing I don't know how to address is to decide if it is better to use the 480V coils or the 55V, clearly biggier section is a good thing, but I don't know if having close to 9 times less turns could be an issue.
EDIT: i tried Tiziano's scheme with 480V coils and the feeling is that the efficiency increased, now I get 229V in the (single) output coil