Due to a conflict in OVMS with CLI options and the config file, we can't pass --target_device=NVIDIA as a CLI option. Instead, we need to modify the adapter to pass this argument in the model config file for each model. This code should be conditional on the pod requesting an Nvidia GPU, and behind a flag that can remove it once the AUTO, HETERO, or MULTI plugins can use the Nvidia plugin.

Acceptance Criteria: ModelMesh OVMS adapter is able to pass target_device: NVIDIA argument to OVMS via the config file only when an Nvidia GPU is requested by the pod