Thanks for writing this up, I will certainly have a look to see whether the methods/workarounds you present here could be used in my research, though of course I run most computationally expensive calculations in FORM.
I think your conclusions nicely summarize the feelings of the theoretical physics community; we spend a lot of money on Mathematica licenses, while the development efforts and direction hardly benefit us at all (here I acknowledge at least recent efforts to improve polynomial arithmetic performance through the use of FLINT). It is quite discouraging to see some of the new features introduced lately when some core functionality is so inadequate for our needs. If Mathematica 15 were simply "Mathematica 14, but much faster, and no regressions in Series" I think we would consider this to be a release actually worth paying for.
I would also be very interested to hear any opinions of the developers on this topic.
