If the routine is implemented correctly there's no reason for having a different sound experience but again you've to understand the replay routine first to be able to make it work as the original. No HW limitation here is what I meant.
Yes, but it does not need to if its not played real time, the sound samples might be mixed in a buffer before playback. In the same way you paint a picture and plot the pixel in correct x,y locations, and then view it.
This is not how the original player is playing the sound. Due to the DMA there is variation when old sample stops and new sample starts playing and there is variation when effects are applied to the sound. On a faster machine player routine is executed faster and you may hear subtle difference in the sound compared to a slower machine.
Player routines are more complex than you think. It is not purely "digital".