This is exactly what I was telling you before... axoloti is a 168mhz arm processor (see here) , not a PC with 3 ghz with 8 cores... but its also 60 euros.
on the flip size, it runs a real time OS, dedicated to one task only - your patch.
how many voices you get will be totally dependent on how much process you do... and also how well you implement it.
if you put the resonator per voice, it will consume considerably more cpu, than if you use one for the whole patch.
this is why you need to start small, build things up bit by bit, see how cpu your using, see where you can accept limits, see where you are willing to spend the cpu.
(the thread you mention, my initial look at the patch, it seemed sub-optimal, ive not looked at it since though - I'm sure others are helping improve it, but I personally dont have time to look at all patches/objects created by the community... gut feeling is 8 voices is a reasonable goal, but depends on what fx you do, which probably consumes more cpu)
this is the problem with 'high level thinking'... I can create a patch that will on paper do as you, but if it would meet you expectations.. Ive no idea.
I bought multiple axo boards from the outset, because I felt that with my expectations were high.... so I planned for having multiple voice boards, which would then feed into a dedicate board for 'global fx'
I dont think you should necessarily do the same, I think a more 'sane' route is to get one board, experiment with it, see what you can get from it, try each part of your 'patch' separate, see its cpu usage... then work out how many boards you would need... and then decide if thats what you want.
multiple boards, currently this can be done, using USB/analog outputs into a mixer... this works ok, once you are building your final setup, but its not ideal during development.
there is a plan to have digital audio between boards, but its early days... and by definition will create some latency. (though the good news is, with axo being so low latency, this extra, Id hope will not be an issue.
(though as always, buy for what it can do now... not what it might be able to do in the future!)
of course, I do accept one thing here... the cost of this project, will not really come down to hardware costs , even if you buy 5 boards (not saying you need this), its still what , 300 euro, which is 'cheap'
the real cost, is the investment of time... which, as Ive highlighted before, may be high, depending upon your skill set... and (again repeating myself) , I suspect, in practice, you will find it hard to 'sub contract' work... it may be your interests lie where others do in which case it will work out... but no-one can promise you this, no-one will commit to getting your project to work... they have their own to get on with.