Dont forget Archimedes principle.
You have to subtract the "weight" of the water displaced by the remaining engine, so that the net effect is probably not that much, and is relatively close to the centre of displacement.
End result, not much roll moment.
I'd discount this for the engines. After all, the detached #1 engine sank immediately - thus no appreciable bouyancy. The main airflow path for the fan and core flood immediately; other odd cavities such as rotor interiors take longer. And the turbine disks are nickel-rich alloys - denser than steel.
Immediately after ditching, the ship remained nearly wings level for most of the evacuation, but slowly rolled toward the #2 engine as odd cavities in the right wing slowly flooded. The left wing was thus elevated above the waterline, and never flooded.
By the time it was towed to the Manhattan side, the roll was 45 degrees or so, and remained so until it was slowly hoisted clear of the water.