I think that the memory restriction on plocks has more to do with the number of different parameters you p-lock than the number of triggers you p-lock the same parameter on …
So if I have a pattern and I plock 64 different parameters, that’s going to use more p-lock memory than if I have a pattern with 64 triggers on one instrument, and I plock the same parameter on each of those triggers.
What this means in practice is that if you run out of memory, you can see if there’s some plocked parameter you don’t really need, (like if you were live recording plocks, and tweaked the decay of one instrument on one trigger, but not by enough that you can hear it), and get rid of that single plock.