Green contexts and inter_sm scripts

[fotstrt]

This PR:

• Introduces Green contexts and creation of streams associated with contexts
• Adds inter-sm interference scripts
• Unifies the gpu_model_runner.py for both intra- and inter-sm interference

[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.

🚀

[elvingerpaul]

Looking good, thank you! Mostly about making sure that main.py is consistent with gcontext_test.py. I'm re-running the measurements to check that they overlap with the previous config. Once they match I'll approve the PR :)

One question, is there a specific reason why we don't unify the behavior in having one single interference kernel for the entire decode for inter_sm and intra_sm instead of launching a new decode kernel for each token in the intra_sm case?

[elvingerpaul]

I'll merge the PR, so that I can add code and have all in one place