Yes indeed, the wiring can become overwhelming. For the existing voice assigner, you could do a little proto-board with the new socket and lead-out to a connector arrangement for your Arduino Mega to help clean up.
For sure, I believe the main CPU must be a daunting task. In fact even your work with the Key assigner seems non-trivial to start from scratch.
As for the new oscs, since I do not have as much flexibility as you in the Polysix, i.e. tapping into the Key assigner, which has the very cool side benefit of allowing you to sync any Arduino MIDI voice to the Polysix filter Envelope, I was going to do the same simpler thing that I planned with my Korg DW-8000, i.e. just put 6 separate input jacks into the VCF sections.
For the DW-8000 for now, I have only tried getting into the VCF through the main Noise generator input. This is distributed to all voices, but ultimately, I will do 8 inputs so that I can plug 8 separate set of oscs into the filters. Here, what is fun is that the DW-8000 is already MIDI-enabled, so that if I trigger the external Oscs with my master keyboard, I am also triggering the DW-8000 filter envelopes simultaneously: 8 new osc sets with plenty of flexibility (including Sync, FM, AM, Cross-Mod, etc...) + 8 synchronized VCF from Korg NJM2029s - should be awesome, and my own custom little hybrid.
So for the Polysix, I am planning 6 separate inputs, each with the 33K resistor to each voice. The fact there's no Key trigger sync yet for both in my setup makes it little benefit, so I will need to do either what you did, or get a Kiwisix or some alternative or build my own to reap the full rewards of more oscs into the SSM filters.
Cheers!