imho some nodes could need a vector size pin additional to the bin size pin to save a lot of performance using swapdim and count and swap back to do the necessary calculations:
*all spectral nodes (+, *,mean, rootmeansquare,…)
*CAR, CDR, differential, integral
*resample
(*b-spline)
e.g: a spread with 15 slices could be treated as 5 3dvectors. and you could easily calculate with binsize 2 and 3. making modules for this creates a huge nodecount.
furthermore a vector (join) and vector (split) with vectorsize would also make patching life more convenient. e.g passing position, scaling, and id of an object as 7D vector.
just a thought, just a wish