> R2 should be recalculated as R1+output impedance of the opamp stage.
Exactly correct, but insignificant. The opamp output Z should be awful close to zero. And the tolerance is 10K +/-1% or 100Ω leeway.
I suppose you could have an opamp rise above 10Ω at the top of the audio band, and you might have an input that was better than 1%, and you might actually need "high" CMRR, so you might want to make the correction.
> omiting the C2 cap?
If you care at all, you probably need C2. Note that 47uFd is ~200Ω at the bottom of the audio band, ~60Ω at power line frequency, MUCH more significant down there than any 47Ω resistor and on the edge of my 10K +/-1% (100Ω) criteria. Either use C2 or omit C1 (or make C1 MUCH bigger, 470+uFd).
It works for Chris because his inputs are good enough for his local electrical noise. In fact all this "impedance balanced" stuff is overkill in many-many situations. The box-makers (should) know this, but "can't" give you an unbalanced output (people say "yuck"), so they came up with this 20-cent quasi-Balanced design. (Most boxes don't run to Wima output caps.) And the common 1-opamp "balanced" input really is sensitive to few-ohm source unbalance (and balanced-Z drive is only a partial fix).
> what do you think
What I think is: unbalanced outputs, proper differential inputs, transformers when you cross power grounds (including some PC interfaces: the PC "ground" is nothing like the wall-plug ground). And if that's not good enough, move to a quieter part of the world. Alternatives include improving your inputs (most so-called "balanced" inputs suck).
Proper balanced-DRIVE output have a place in large systems with many different signals in the same conduit. If the voltages on each line are equal-but-opposite, cross-talk is greatly reduced. This is rarely an issue in music production studios: everything in the room will end up in the same groove anyway.