[P/D Disagg][Benchmarking] One request at a time benchmarking for P/D

[tlrmchlsmth]

Benchmarking script and modifications to the toy proxy server to benchmark one request at a time. Since we can't turn off prefix caching for P/D, we reset the prefix caches between requests.

This is a worst-case benchmark for P/D to expose the TTFT overheads and a test to measure and optimize the KV transfer.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

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 either: Add ready label to the PR or enable auto-merge.

🚀