Agreed MFS, surface roughness is a big problem with ice. And yes, the mechanism is as flow separation. Hence many hypersonic missiles have, counterintuitively, a roughened nose surface. This means that the air flow separates and becomes turbulent close to the nose of the missile meaning, again counterintuitively, less drag-production by the rest of the body.
I would still suggest that any significant ice accretion is likely to have the effect of changing, adversely, the lift producing characteristics of an aerofoil section. And the big problem is that this is entierly unpredictable, leading to flight with an unknown envelope which can readily lead to problems.
And yrvld, I seem to remember something to do with reverse thrust being used on the ground to back out of the gate, causing ingestion of large amounts of slush and ice into the engines. This exacerbated by the omission of engine anit-icing use. Of course, to try and isolate one cause is stupid and counterproductive, but it's a very interesting example of the holes lining up!