The most obvious reasons for paralleling any topology that I can think of would be for power and/or impedance considerations.
I think the specific practical reason we see so many modern parallel SE topologies compared to PP is that in order to get more power out of a tube amp you either have to move up to the next class of power tube, or parallel them. Consider that anything north of 500VDC starts to look a little scary in terms of components, safety, etc. At that voltage you can find a lot of PP amps that put out well over 30WPC. A lot of SE amps in that voltage range put out around a dozen. To get more power from SE at those voltages a lot of manufacturers parallel the devices. It makes a lot of sense, especially in the case of tubes like 45 or 2A3 where the non-paralleled version puts out so very few watts as to restrict the real-world compatibility to a narrow speaker class. A lot of people read about the "magic" of a certain tube, want to try it for themselves in SE, and simply can't because their entire system has evolved in a completely different direction up to that point.
This is not to say that paralleling output devices is a cure-all for the power problem. There is a trade off between complexity of circuit, and power gains realized. A simpler circuit will result in the natural imbalances between the tubes disproportionately favoring one tube, which then gets run harder, which exacerbates the problem, and you only realize a fraction of the power gain expected.
A more complex circuit could do all sorts of stuff to smooth over tube matching efforts, but in my opinion at that point you start to move away from the simplicity and purity that single ended amps are known for. If I needed more power from a specific tube, I'd just go and make it push-pull at that point; Then again that's just me.