> I *think* I calculated 281 ohms for the Rs
If your engineering department specified an exact attenuation value, you better calculate.
If you are making music, instead of making meter readings, 220 will work.
> so I don't have to worry about "shorting" the unused channels
In these cases where the desired attenuation is much more than the minimum attenuation, you needn't "worry".
A minimum loss 10-input mixer would have about 20dB loss; with 20K sources, about 2K output impedance.
But when mixing to a mike-amp, we need more like 40dB loss and an output closer to 200 ohms. So we add Rs.
Consider ten 20K legs loaded in 220.
10 legs driven, bus impedance is (20K/10)||220= 2K||220= 198 ohms.
1 leg driven, 9 legs left open: 20K||220= 218 ohms.
Few mike-amps will be upset by a change from 198 to 218 ohms.
Difference in level is roughly 218/198= 1.099 = 0.8dB.
You could short/unshort ALL the unused legs while the audience (or recorder) was listening, and not notice.
(I wouldn't like plug/unplug to live electronics during a critical soft passage because clicks and thumps from small DC offsets would come through, even though the gain change was negligible.)
I would point out that your 4K32+10K0 could just as well be 14K or 15K. 10% tolerance give 1dB errors so I'd use 5% accuracy (more for meter calibration than musical effect). 5% values would allow much better than 26dB CMRR, which is probably not really mixer-limited and probably not a key parameter in a small studio. Given that 1% resistors are available and the cheapest part of the box, they are fine, but I don't see any advantage of doing 10K+4K1 instead of just the nearest single value to 14K1.
