Thanks, for the suggestions. Just for clarification, I'm using 1WE, 1CE, 1RE (or 1WE - 1 CE/RE) for each rodeostat. For now I'm running each rodeostat in individual beakers.
Anecdotally, I ran the same experiment using a research-grade multi-channel potentiostat (from GAMRY) and didn't see cross-talk in a single beaker. Specifically, I ran high currents using a single Ag/AgCl wire (reference+aux) for each channel. The stability of each cell degraded randomly. Previously, when using a few rodeostats, all WE signals decayed at the same time (suggesting the 3 wires were acting as a shared reference/aux).
Ordered a USB isolator, will wait before splicing my teensy boards. They're cheap enough to replace, but I'd rather not need to if I can help it.