Treat failures in automatic CDI spec generation as an error

[elezar]

At present, when automatic CDI spec generation fails, the error is logged and execution continues handling the automatic CDI request as a regular CDI request. This leads to cryptic errors such as:

sudo docker run --rm --runtime=nvidia --gpus all ubuntu nvidia-smi
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: could not apply required modification to OCI specification: error modifying OCI spec: failed to inject CDI devices: unresolvable CDI de          vices runtime.nvidia.com/gpu=all: unknown

We should rather return the error from automatic spec generation directly to surface errors to the user.