Cherry pick Profiler bug commits

[shenoyvvarun]

Cherry pick
vllm-project/vllm#21798
vllm-project/vllm#21842

[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.

🚀

[gemini-code-assist]

Code Review

This pull request cherry-picks two bug fixes related to the multimodal profiler. The core change correctly calculates the maximum number of tokens for multimodal inputs by including non-embedding tokens, which is a crucial fix for accurate memory profiling. A new test case for mllama4 has been added to verify this new logic.

The changes are generally well-implemented, but I've identified a critical issue: a leftover merge conflict marker that will break the build. Additionally, there's a missing return type hint in a new public method, which should be addressed to maintain code quality.

Once these issues are resolved, the pull request should be in good shape.