Hello,
from a user perspective, the most flexible option IMHO is to support multichannel both in the Cabbage player and in the VST export, and use the actual number of channels that the effect/synth is using. For stereo effects, export a vst with 2_in/2_out, for other configurations, export N_in/M_out. The rationale for this is “CSound fully supports multichannel” and “Cabbage is a frontend to CSound” therefore “Cabbage fully supports multichannel”. This is what I thought in the first place when I started to develop a multichannel effect in Cabbage.
Having said that, this would be of course the ideal, but I don’t known nothing about how difficult this would be to implement and to maintain in the future, how time/effort needs to be put into it and which are your priorities for the project evolution. So this is only my (uninformed) opinion.