I respectfully disagree with your input voltage calculations. I’m not sure how you arrived at 17.1VDC input voltage…
To calculate AC RMS voltage to DC volts you multiply the AC RMS volts by the square root of 2 (which is 1.414). And keep in mind while the voltage from the mains transformer secondary to the center tap is documented as 19VAC, the swing from secondary to secondary is then 38VAC. That 38VAC to rectified raw DC then is about 53.7VDC. And this assumes your mains voltage is the rated standard. The schematic assumes my mains voltage is 117VAC when in fact it’s usually 121-123VAC, sometimes even 124VAC. So that means, in my case, the rectified raw DC could be as much as 57VDC. Yes there is some voltage drop across the bridge, and there may be minimal voltage drop across the main filter caps, but with the total drop across both sides of the bipolar input, it is still probably about 52VDC. I could measure but can’t easily open my JUNO-106 up right now. So what that means though is the Vcc+ and Vcc- inputs are each seeing, in theory, about 26VDC. So that’s what could pass to downstream systems if the regulator fails…and yes that all depends on how it fails and I don’t know the common failure modes for the M5230L. But regardless it seems like super cheap insurance to put the zeners in and ensure the maximum output from the +/-15V supply is +/-17V.