Fix reasoning prompt injection in EvoAgentBench reasoning adapter

[CZH-THU]

Summary

Update omnimath.py so reasoning prompt injection only appends skill_text when it is actually present. If skill_text is empty or absent, the original prompt is returned unchanged.

Area

• Architecture method
• Benchmark
• Use case
• Documentation
• Developer experience
• CI, build, or release

Verification

git status
git diff --stat

Checklist

• I kept the change scoped to the relevant area.
• I updated docs, examples, or setup notes when behavior changed.
• I added or updated tests when the change affects behavior.
• I did not commit secrets, .env files, dependency folders, or generated output.
• Active relative links in Markdown files resolve.

Notes for Reviewers

• Focus on omnimath.py
• The patch ensures skill_text injection follows the same conditional behavior expected from the reasoning task_field flow.