Hi lhwidget,
many thanks for sharing your experience.
The fact you aren't hearing noise/pop/clicks when stopping a song and playing another one has a plausible explanation:
your card has muting at the analog output that is used to cut sound for a fraction of second after SPIDF synchronization.
As you can see from the attached pic, the tiny transistors (Q2-Q3-Q4-Q5) in the upper left of the EMU1212 image are very likely the electronic "relays" that mute the outputs.
Not all DACs in the market implement that muting function as many audiophiles believe that those circuits might cause sound degradation.
Now, I don't want to enter in the merit of the choices behind muting implementation but the fact is, if no muting is adopted, everything that comes out from the PC is converted to sound, glitches included.
The fact that some DACs implement mute and some not might explain why the "click" problem isn't so widespread (beside the fact that some people may not consider it a problem).
I am almost convinced that
MC is doing what it is supposed to do and does it properly. However, although not causing it directly, MC could
help circumventing the problem by implementing the simple measure I've proposed earlier and my reasoning is simple: if clicks are proportional to intensity of sound before stop is pressed, allowing through an option to stream a user selectable duration of silence just after stop button is pressed should eliminate the problem before it is generated.