From the report, I'd imagine he was unaware the left engine was failing as the CSU was masking its surging to some extent. He'd be aware 'something' wasn't right as the aircraft departed heading, hence the go-around being initiated. Then I think what followed would happen too quickly for him to do much about it.
Incidentally did I read the report correctly re the anti-feather latches? If the engine fails in that the oil pressure to the CSU drops, the prop will feather, but if the engine is stopped on the ground in the normal way, the latches hold the blades at positive pitch? And in this case, where the engine ran down due lack of fuel the prop would behave as if in a ground shut shut-down; the latches would engage and the prop would not therefore have feathered?
If so, that seems an unwise design flaw, doesn't it? Surely one always wants the prop on a twin to feather if its engine fails?