nvcr container & backend hosting question

[here4dadata]

The nvcr.io/nvidia/tensorrt-llm/release:0.20.0 container does not include the triton_backend/ folder, is this the norm going forward for future containers? Should these templates just be copied out of the github project sep. before creating a repos folder for further triton backend hosting?

Is the guidance to also use llm_api/tensorrt_llm going forward, or is there a future for the inflight_batcher_llm and dissaggregated serving c++ methods?

Thanks!