Summary: like you, I'm sceptical, but it is plausible.
In the absence of any other differences between the panels, you're right that it doesn't make sense that an array of twelve 250W panels is more efficient than an array of ten 300W panels. Given that it's the same brand of panel, the same efficiency of panel, and the same type of panel - polycrystalline - it's very hard to see why 12 panels would be more efficient.
It's conceivable that your supplier can just get the 250W panels cheaper, and so is trying to nudge you in that direction. Or, if some part of the array might be subject to shading during part of the day, perhaps the array of twelve panels might just generate more power when there is partial shading.
Or it could be that there's the inverter operates with different efficiency depending on the input voltage; the input voltage is driven by the number of panels you have in series (assuming the panels have the same voltage, and the 300W panel just gives a higher current). Twelve panels gives the option of installing 1,2,3, 4, 6 or 12 in series; whereas an array of 10 panels could be 1, 2, 5 or 10 in series. The inverter might operate more efficiently with one of the formations unique to a 12-panel array (3, 6 or 12 in series) than any of the formations available to a 10-panel array. This would not apply in the case of say, the Suntech panels, where the 300W panel seems (according to my reading of the datasheet) to scale up voltage rather than current, so 12 250W Suntech panels in series have the same voltage as 10 300W Suntech panels.