there are lots of things you can do
you do not have to use every input and output of a module
you can/should mix outputs from a vco before filtering
do you need to mix all outputs together - no
can you process waveforms differently and then mix before filtering - yes send the non sub through some saturation or a delay or a pitch shifter - and then recombine with the sub before filtering perhaps
can you process outputs separately through different processing chains - yes
can you mix waveforms from different oscillators together before filtering them as one - yes
do vca/vcf/eg etc etc need to be the same model - no - they can be whatever you want them to be - how they behave is how you personalise your modular synth
this is why: sound sources < sound modifiers < modulation sources < utilities works perfectly - for example:
a sound source may have multiple outputs (different waveforms + sub) which want to be mixed (utility - maybe to 2 outputs) and then filtered (sound modifier - maybe 2 filters) and then mixed back through a cascading vca (utility) which needs envelopes (modulation source) - plus somewhere in there there are a lot modulation inputs some of which will want an lfos and/or other modulation sources combined in another (matrix) mixer (utility) perhaps etc etc etc
the more mixers, vcas, mults, envelope generators (and other modulation sources in general) you have the more choices you have in the way you patch your modular together - which makes for more interesting patches
it's fine to have 12 different vcas or 12 identical vcas (or envelope generators or filters or whatever) - it's up to you - generally a selection of linear and exponential (or ones that do both) is good to have though
"some of the best base-level info to remember can be found in Jim's sigfile" @Lugia
Utility modules are the dull polish that makes the shiny modules actually shine!!!
sound sources < sound modifiers < modulation sources < utilities