[fix](test) stabilize sql digest generation case

[shuke987]

Summary

• Use a deterministic slow query in test_sql_digest_generation so threshold 0 reliably satisfies the FE condition elapseMs > sql_digest_generation_threshold_ms.
• Apply the same query in both low-threshold and high-threshold branches to keep the assertion target consistent.

Testing

• git diff --check
• Not run locally; regression environment not available in this session.

[shuke987]

run buildall

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[shuke987]

/review

[github-actions]

Automated review completed for PR #65343.

Overall opinion: no blocking issue found. I reviewed the single changed regression test and the relevant audit/sql-digest/sleep-function paths. The change makes the audited query deterministically exceed sql_digest_generation_threshold_ms = 0 while still remaining far below the high-threshold branch, and I did not find a substantiated correctness or compatibility issue to raise inline.

Critical checkpoints:

• Changed files reviewed: regression-test/suites/audit/test_sql_digest_generation.groovy.
• Existing inline review context: no existing inline comments or replies were present, so no duplicate issue was suppressed from prior threads.
• User focus: no additional user-provided focus was supplied.
• Optimizer/rewrite correctness: sleep(1) is a registered Nereids scalar function and the BE implementation avoids constant-time open-stage execution; no semantic rewrite issue found.
• Tests/session/config behavior: setFeConfigTemporary restores FE config, the low/high threshold assertions remain aligned with AuditLogHelper's elapseMs > sql_digest_generation_threshold_ms check, and no expected-output or style issue was found.
• Validation limits: static review only; the regression suite was not run locally in this runner. git show --check was clean. Current GitHub checks include passing style/lightweight checks, with TeamCity compile and macOS BE UT still pending at review time.

Subagent conclusions:

• optimizer-rewrite: no candidates appended; convergence round 1 returned NO_NEW_VALUABLE_FINDINGS.
• tests-session-config: no candidates appended; convergence round 1 returned NO_NEW_VALUABLE_FINDINGS.
• No subagent candidate became an inline comment, and there were no duplicate candidate merges. Convergence ended in round 1 with all live subagents reporting NO_NEW_VALUABLE_FINDINGS for the same empty proposed final inline comment set.

[shuke987]

run buildall