I fully agree with this, however I think there is a good point buried in the post.
In helix we have a multi-output block that ties the output of the modeller to a single volume control knob, and then ties the output of the volume knob to the output jacks.
The AES output jack simply takes the digital signal from the output of the volume knob and sends it. This signal has already been transposed by the volume knob (i.e. all 24-bit numbers adjusted up or down by the same amount).
The XLR output takes the same digital signal but has to convert it to an analogue signal prior to sending it. There is no fixed way to convert the digital signals volume, however - it can be mapped to anything that conforms to the standard for that analogue signal. There's a lot of wiggle room there (but even at this point, there is no inherent "volume" in the signal. Both the analogue and the digital signal are only encoding differences in volume).
So we now send both a digital and a converted analogue signal to the same device - a powercab in this example. The powercab does the reverse operation on the XLR signal and converts it to digital. Again it has a choice about how 'loud' it interprets the XLR signal in digital terms. It then does it's modelling thing and the modified signal is sent to the speaker for another round of analogue conversion in the form of moving a speaker. For the digital input it does nothing before modelling (assuming no input gain change), no conversion required.
When the signal hits the speaker is the first time there's an inherent volume to the signal. Prior to this we were only encoding differences in volume.
Now, all things being equal - all the volume settings on the helix, and all volume settings on the powercab - we SHOULD get roughly the same volume at the speaker whether we use the AES output or the XLR output on the Helix. Not for a technical reason but for a practical reason. But we don't.
It doesn't make sense to have different signal transport methods produce wildly different results with all other things being equal. I agree with OP to an extent - I wouldn't expect the volumes to be totally matched but I wouldn't expect them to be totally different either (say by more than 6dB).
So is this the fault of the helix or the powercab? It could be either, it's not possible to say as there are two places the signal is being converted where there's a somewhat arbitrary decision about how to translate a signal's volume from one form to another - before it's sent by the Helix and after it's received by the Powercab.
But certainly it's fixable in the firmware of either.
It's also totally fixable by the user. By adjusting the 'Digital Output Level' setting up or down to match to the level of Helix-XLR + Powercab-XLR conversion. Unlike all other volume controls, this setting will ONLY affect the AES output. I think this is the wrong way of thinking about however, I view this setting more as on offset for the XLR output, where there is some interpretation of volume going on.
In short, I agree with OP that there's a questionable aspect of the implementation, but not that it needs to be fixed by Line 6. You figure out the difference once, you adjust the setting, then you forget about it. Now you can switch between digital and analogue outputs to your heart's content.