[AutoDeploy]: Investigate streaming runtime knobs and apply where applicable

🚀 The feature, motivation and pitch

Follow up on https://github.com/NVIDIA/TensorRT-LLM/pull/14943
Investigate meaning and impact of

        self.llm_args.stream_interval
         self.llm_args.batch_wait_timeout_ms
        self.llm_args.batch_wait_timeout_iters
        self.llm_args.batch_wait_max_tokens_ratio

When should we set each of these to non-default values?
FYI - in examples/configs/curated/ only stream_interval is ever set. The others might be less useful, but we should still try to understand their purpose.
stream_interval was already discovered to be beneficial for gpt-oss and llama3.1.
Odds are it will also be beneficial for nanov3 - this is a good place to start with (quick win).

Done criteria:

• Understand meaning of each of the above knobs
• Measure impact and tune to optimal values
• Issue PR to update configs for all applicable BTK models (update model registry configs)

Alternatives

No response

Additional context

No response

Before submitting a new issue...

• Make sure you already searched for relevant issues, and checked the documentation and examples for answers to frequently asked questions.