Hello Sebhuber,
Diffraction is wider when the diameter of the hole is smaller, but all audio frequecies are diffracted when the diameter is 1 cm or smaller. From a 90° angle, a 17 kHz frequency shows one alternance of its wave through a 1 cm gap. This is not enough to cancel the diffraction. Lower frequencies are even less canceled, and smaller diameters cancel less also.
So any IEM will diffract all frequencies through its output.
Reflexions may take part in the soundstage, but the distances involved with IEM seem extremely small. Floyd Toole shows soundstage effects occuring with reflexions delayed by several ms to several tens of ms. The delays of the reflexions in such small spaces are at least ten times shorter, and Toole's detection curves start to fall down below 8 ms. Therefore the effect of reflections in IEMs must be small. I don't know if it is null, though.
The frequency response may be a bigger factor for soundstage. It can be affected by the diameter of the IEM, I suppose. I would think that if the diameter affects soundstage, it would be through frequency response. But that's just a guess.