That's almost kind of totally not quite right. Unless a program has something like DX12's heterogenous explicit mGPU baked in, then programs don't split resources like you are thinking. They just use the GPU that Windows, the optimus driver, etc. tells it to use. In the past a secondary nvidia card could be added to a desktop to do hardware physx calculations, but now (at least for games, I don't know about other use cases) physx is almost exclusively done in software on the CPU because hardware physx doesn't run on AMD and Intel GPUs. Background tasking, maybe, but how that is divided can depend on things like whether external displays are being used. If stuff in the background is crazy enough for this to be relevant it's time to close some programs.With dgpu active, the igpu takes a secondary role, supporting the dgpu by doing extra stuff like physX calculations, background tasking etc. The load is somewhat split between the 2 with the dgpu doing the lions share of any 3d rendering. Disabling the igpu can have a negative impact as that then forces the dgpu to do all the work, and if not plugged in, will be further handicapped by power limits.
Disabling the iGPU tends to help performance (or be neutral at worst) as far as gaming goes on these low voltage comps. The iGPU and CPU share the power budget for the CPU package (dGPU power consumption, however, is not counted against this). If the iGPU is turned off, it's not generating heat, drawing power or using system resources, so there's that much more of the power and thermal budgets available to the CPU cores.