doesn’t make a ton of sense to me.
norns scripting layer is a couple of things: 1) wrapper around hardware, 2) launcher / UI for user-created musical logic. those are singleton components by their nature. executing parallel musical logics (multiple lua VMs or something) poses challenges that are not directly related to compute resource allocation and are more about a) UX, b) managing parallel processes.
in other words: norns “scripts” are lua applications that can nest any number of abstraction; norns “engines” are SC classes or standalone DSP programs that can likewise nest any level of modularity.
you see norns “library” offerings just doing one thing. but IMHO that’s the nature of things that are shared. if i mash together several functionalities in one engine (which is super easy to do if you have some SC fluency) it’s not gonna be necessarily useful to a wide range of people with different creative needs. (e.g. i made an engine to play the set for the tour i’m currently on. it does three arbitrary things. it’s not a “tool,” it’s a piece. i do software architecture for my day job, not for my musical practice.)
the audio processing “guts” mostly provide SuperCollider classes for wrapping user code and talking to scripting layer. SC is already a rich development environment. i dunno what you mean by “multitracked.” if it’s polyphony/polytimbrality, that’s there if you want it already. if you mean more audio I/O, that’s a hardware limitation that is unrelated to cpu/ram/&c.
there is also a built-in effect chain in norns, that runs on a separate processing thread from SC and doesn’t directly compete for CPU resources. this stuff is hardcoded, but it can be extended to something like an LV2 host chain any time anyone wants to implement that. (i’ll help, but i can’t take it on.)
anyways, more processing power would allow you to do more in each of those environments. that’s awesome. the R engine is the norns modular and more CPU would let you instance more modules; nice. but i don’t see a qualitative difference, and there are plenty of optimizations remaining w/r/t how those environments make use of existing resources in CM3(+).