Force use C++17 globally to avoid compilation error

[chenfengjin]

Purpose

vLLM use some c++14/c++ 17 features, which may result in compilation error.

• std::decay_t used in scaler_types.hpp ,required c++14 or above
• std::optional in many files, which required c++ 17 or above

Force use 17 as default CXX version may reduce many confusing errors when build from source on some platforms that use C++ 11 as default CXX standard like MacOS apple silicon

Test Plan

Test Result

---

Essential Elements of an Effective PR Description Checklist

• The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
• The test plan, such as providing test command.
• The test results, such as pasting the results comparison before and after, or e2e results
• (Optional) The necessary documentation update, such as updating supported_models.md and examples for a new model.
• (Optional) Release notes update. If your change is user facing, please update the release notes draft in the Google Doc.

[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 ask your reviewers to trigger select CI tests on top of fastcheck CI.

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.

If you have any questions, please reach out to us on Slack at https://slack.vllm.ai.

🚀

[pytorch-bot]

No ciflow labels are configured for this repo.
For information on how to enable CIFlow bot see this wiki

[gemini-code-assist]

Code Review

This pull request correctly enforces the C++17 standard across the project by setting CMAKE_CXX_STANDARD and CMAKE_CXX_STANDARD_REQUIRED in the root CMakeLists.txt. This is a good practice for ensuring build consistency. However, this change introduces a conflict with an existing per-target C++ standard setting, which could lead to maintenance issues in the future. I've added a comment with details on how to resolve this.

[pytorch-bot]

No ciflow labels are configured for this repo.
For information on how to enable CIFlow bot see this wiki

[pytorch-bot]

No ciflow labels are configured for this repo.
For information on how to enable CIFlow bot see this wiki

[pytorch-bot]

No ciflow labels are configured for this repo.
For information on how to enable CIFlow bot see this wiki

[cjackal]

Interestingly mergify bot and pytorch bot are battling here 🤣

[pytorch-bot]

No ciflow labels are configured for this repo.
For information on how to enable CIFlow bot see this wiki

[LucasWilkinson]

LGTM; thanks for the contribution!