update llama triton example

[zhyncs]

Motivation

After seeing #1101, I verified it in my local development environment and it ran normally. At the same time, I found that this example was not easy to use, so I added some comments and synchronized start_ids.csv from FasterTransformer.

Modification

as titled

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness.
3. If the modification has a dependency on downstream projects of a newer version, this PR should be tested with all supported versions of downstream projects.
4. The documentation has been modified accordingly, like docstring or example tutorials.

[zhyncs]

e.g.

CUDA_VISIBLE_DEVICES=0 ./bld/bin/llama_triton_example /workdir/lmdeploy/examples/cpp/llama/llama_config.ini /workdir/lmdeploy/examples/cpp/llama/start_ids.csv

[zhyncs]

Hi @lzhangzz May you help review this pr? Thanks.