In this case there AudioSource is there merely to schedule blocks of audio to be processed. The output of the AudioSource is muted. You;ll see this in the OnAudioFilterRead() method.
Ah, I get it now. So in this case you have two options. 1) Continue to use an AudioSource components, merely to schedule blocks of audio to be sent to Csound, or 2) set up a performance thread to run an instance of Csound. 1 can be implemented without any changes to the current scripts. 2 would involve a little bit of a rewrite of the C# stuff and how Csound is called. But reading multichannel audio is simple in Csound. This little bit of code will read all 32 channels of that file, and output the first two:
aSignal diskin2 "C:\\Users\\rory\\Downloads\\32ch-96000-rf64\\32ch-96000-rf64.wav", 1, 0, 1
outs aSignal, aSignal
Btw, I assume these aren't one-shot samples you're looking to fire off in quick succession? I imagine performance might be a little sluggish if you're using 32 channels of audio for a single gunshot for example!