I feel like I should chip in on this subject, since I have spent quite some time optimizing the pitch shifters. Before the optimization process nearly all of them were unusable because of very high CPU-loads. I iteratively benchmarked the code of the plugins to find the bottlenecks, and then rewrote parts of the code to make it run more efficiently.
The potential for further optimizing these plugins (on a C++ level) is limited. The truly big CPU-consumers are already taken out of the equation.
From the research that I have done, it should still be possible to optimize the plugins on an Assembly level. Theoretically it should be possible to drop the CPU-load down to about a third of the current load, but this would take an extreme amount of time (months).
Currently the MOD team is not working on plugin alternatives to the currently available pitch-shifters and harmonizers. The gen~ integration is definitely interesting for this subject, but I did not personally do any tests with regard to CPU-load.
If anybody is interested in further optimizing the pitch-shifter family, we are open to pull requests!
I hope that answered most of your questions.