[Misc] Add tqdm progress bar during graph capture

[mgoin]

Nice idea inspired by sgl-project/sglang#2502

Example of log:

vllm serve meta-llama/Llama-3.2-1B-Instruct
 ...
INFO 12-19 22:58:27 model_runner.py:1415] Capturing cudagraphs for decoding. This may lead to unexpected consequences if the model is not static. To run the model in eager mode, set 'enforce_eager=True' or use '--enforce-eager' in the CLI. If out-of-memory error occurs during cudagraph capture, consider decreasing `gpu_memory_utilization` or switching to eager mode. You can also reduce the `max_num_seqs` as needed to decrease memory usage.
Capturing CUDA graph shapes: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 35/35 [00:13<00:00,  2.51it/s]
INFO 12-19 22:58:41 model_runner.py:1535] Graph capturing finished in 14 secs, took 0.21 GiB

vllm serve meta-llama/Llama-3.2-1B-Instruct -tp 2
 ...
INFO 12-19 22:57:03 model_runner.py:1415] Capturing cudagraphs for decoding. This may lead to unexpected consequences if the model is not static. To run the model in eager mode, set 'enforce_eager=True' or use '--enforce-eager' in the CLI. If out-of-memory error occurs during cudagraph capture, consider decreasing `gpu_memory_utilization` or switching to eager mode. You can also reduce the `max_num_seqs` as needed to decrease memory usage.
Capturing CUDA graph shapes:  80%|██████████████████████████████████████████████████████████████████████████████████████████████▍                       | 28/35 [00:13<00:02,  2.64it/s](VllmWorkerProcess pid=2099548) INFO 12-19 22:57:17 custom_all_reduce.py:224] Registering 1155 cuda graph addresses
Capturing CUDA graph shapes: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 35/35 [00:15<00:00,  2.20it/s]
INFO 12-19 22:57:19 custom_all_reduce.py:224] Registering 1155 cuda graph addresses
INFO 12-19 22:57:19 model_runner.py:1535] Graph capturing finished in 16 secs, took 0.14 GiB

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

• Add ready label to the PR
• Enable auto-merge.

🚀

[jeejeelee]

It's very useful! thanks

INFO 12-20 01:21:29 model_runner.py:1415] Capturing cudagraphs for decoding. This may lead to unexpected consequences if the model is not static. To run the model in eager mode, set 'enforce_eager=True' or use '--enforce-eager' in the CLI. If out-of-memory error occurs during cudagraph capture, consider decreasing `gpu_memory_utilization` or switching to eager mode. You can also reduce the `max_num_seqs` as needed to decrease memory usage.
Capturing CUDA graph shapes: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:06<00:00,  1.32s/it]
INFO 12-20 01:21:36 model_runner.py:1535] Graph capturing finished in 7 secs, took 0.09 GiB