Very nice results that challenge the conventional wisdom. DNN can work with symbolic data and calculations. I think that, it says a lot about our poor theoretical understanding of deep neural networks despite all the practical advances.
I am not trying to underestimate the value of the work but those problems are very much what DNN are trained to solve, are they not?
The fact that they are "symbolic" is just an artifice of the problem: they are no more symbolic than a series of colors, if you tell me.
That is why they can be automated (mostly) and why "tricks" to solve families of them can be developed.