Regarding the FM mod and its ability to track the pitch of the keyboard...
To have any chance of tracking, you need to get the CV for the voices. You only have one LFO, so you'll only be able to track the pitch of one of the voices (hence my comment about putting the synth in unison mode, or in single-note chord mem mode).
The real trick is whether you want to get the pitch CV while it's in Volt/Oct scaling (ie, before the exponential circuit) or when it's in Volt/Hz scaling (ie, after the exponential circuit). I'm not sure which gives better FM effects.
If you want the Volt/Oct scaling, you're gonna be in trouble because the Polysix keeps the pitch CV for all six voices multiplexed (in time) along with a bunch of other control signals on a single wire. So, if you really want this pitch CV, you'll have to listen to the three clock signals and grab the pitch CV (with your Arduino) at just those times when the clock lines are set correctly. This is feasible to do with an Arduino (I think) and I can help with what High/Low states you need to watch for the clock lines, but it will not be a super-easy thing to do.
Much easier would be to grab the pitch CV after the expo circuit when it is in Volt/Hz scaling. These Volt/Hz pitch CVs are already demultiplexed and just sitting on a wire waiting to be grabbed whenever you feel like it...no special timing necessary.
Chip