Sometimes I don't know why the samples are recorded dry and not directly with the MIR-sound....
Even if a person does everything inside MIR, and always with non-creative MIR settings, it's still better to start with dry samples. When I'm working with non-VSL libraries, where there's reverb in the samples, I can hear the reverb getting cut off unnaturally when the note is released. This is most evident in series of fast staccatos, where the amount of perceived reverb fluctuates awfully. And the legatos sound bad too, because as the samples are crossfaded, the reverb of the two samples clashes dissonantly. For this reason alone, it's much, much better to record dry and add reverb only after the attack, release, and crossfade have been applied to the samples.
Personally, I produce music in all sorts of unnatural spaces, and again I want dry samples here so I can process them however I want before adding any reverb. But I'm still most interested in MIR PRO, because I'm demoing MIRx, and for those times I do want to simulate a natural space, MIRx is a lot more skilled than I am.