build: Upgrade to 24.07, TRT-LLM 0.11.0, and Triton CLI v0.0.10

[rmccorm4]

• Bumps Triton version to 24.07
• Bumps TRT-LLM version to 0.11.0
  • Updates model config templates
  • Updated template parser to account for new fields and move more sensible defaults into the parser rather than remembering to modify the templates each release
• Bumps Triton CLI version to 0.0.10
• Adds minor improvement to triton infer to include inputs in output for easier dev/debugging

---

Tested gpt2 locally in 24.07 TRT-LLM container. Will see if llama/opt models pass CI pipeline.

[KrishnanPrash]

LGTM!

[rmccorm4]

Tested TRT-LLM locally for sanity check while pipelines are running:

export IMAGE_KIND=TRTLLM
export TRTLLM_MODEL="llama-3-8b-instruct"
pytest -s -v

=== 51 passed, 4 skipped in 123.80s (0:02:03) ===

[rmccorm4]

Pipeline 17230346 passed. If the linked PR triggered by this PR fails, it has been due to some strange node allocation timeout issues.