Both options will be supported - this allows maxim flexibility. as we are having to go thought the effort of adding a second XMOS for USB to I2S conversion I dont really see the advantage of the direct I2S especially if its limited to 192KHz or maybe 384KHz. The MDAC2 will support 768KHz / DSD512 which is possible over USB2 (not sure if this is the case with the RPi).
The C3 will be mounted on a daughter PCB which then connects to the MDAC2 Mainboard - the best chance of CM3 upgrade in the future is if RPi introduced say a CM4 compute module pin compatible with the CM3 as they have done with the CM1.
This is a Second independent USB2.0 channel otherwise if we shared a single USB channel and switched between internal RPi (SOC) or external USB host we would need to keep enumerating as we switched inputs - and thats a complete pain!
Yes and no - the Detox does bring an improvement to the external USB device and while we will have a "baby" Detox on the SOC USB channel its not the full implementation of the external Detox - also I don't like the idea of the SOC radiating inside the MDAC2...
The SOC can be completely powered down - the trouble with the RPi is that I cannot Sync the SOC master clock to the MDAC2's internal clock - the optimised H3 board would have been much better from an RF perspective