Wouldn't defaulting to a 360 degree annular detector for boundary fluorescence corrections actually both over and under estimate the effect of the boundary fluorescence, dependent on the real world geometry and measurement? Assuming you actually managed to convince the user that this is a problem (...) and were attempting to subtract the fluorescence from real data.
For example, assuming a boundary fluorescence issue rather than an inclusion, the Ti La and O Ka measured X-ray intensity would vary dependent on whether the generated X-rays pass through the rutile or the pyrope on their way to the detector, which is dependent on the geometry of the boundary and azimuth of the detector used.
And then if you were measuring traces on multiple detectors and summing the counts... argh!
An interesting thought experiment, but would you be able to put specific elements on specific simulated detectors (or multiple) and get PENEPMA to work out whether it would travel through a particular phase?
I guess this is only going to become more of a problem as we work towards lower accelerating voltages and consequently lower energy emission lines.