[Performance][gpt-oss] Revert gpt-oss max cudagraph size to 1024

[mmangkad]

Purpose

Now that #27840 is merged, can we also revert gpt-oss max cudagraph size to 1024? Basically reverts #25508.

cc @mgoin

Test Plan

Tested

Test Result

Works

---

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.

[gemini-code-assist]

Code Review

This pull request reverts the max_cudagraph_capture_size for gpt-oss from 992 back to 1024, which is a good performance improvement. The changes are correct, including the updated calculation for the number of CUDA graphs. I have one suggestion to improve code maintainability by removing an outdated FIXME comment.

[mgoin]

Okay this should be reasonable if the restriction is lifted now, thanks! cc @nvpohanh for the gpt-oss config update