Better doc for different between timeout and client_timeout of grpc_client.infer

[ShuaiShao93]

Is your feature request related to a problem? Please describe.
In the doc, it says for timeout, it's only only respected by the model that is configured with dynamic batching, and the server can take model-specific actions. But it's not clear what model-specific actions are. Does it terminate the request for all models? How is this different from client_timeout?

Describe the solution you'd like
Update the doc with the details below:

• What are model-specific actions?
• Does client_timeout also terminate all requests?
• Why is timeout only respected by the model that is configured with dynamic batching?