feat: add LLM API as an backend option for Triton CLI

[richardhuo-nv]

1. add LLM API backend option, which require Tensorrt-LLM 0.18.0+
2. Update Tensorrt LLM templates to reflect the latest update in Tensorrt-LLM 0.18.0+
3. Add LLM API templates from the LLM API MR
4. Add the LLM API model detection logic with "having python backend with model.json"

Will merge only if Tensorrt-LLM 0.18.0 gets deployed into triton container.