[fix](regression) Batch export_p0 generated VALUES inserts

[shuke987]

Summary

• Split large generated VALUES setup inserts in export_p0 into 500-row batches.
• Keep the same data volume and OUTFILE validation coverage.
• Avoid single huge literal INSERT statements that can exceed the Cloud P0 send-fragments RPC deadline during fragment preparation.

Related: DORIS-26804

Testing

• git diff --check
• groovyc syntax compile for the three changed Groovy suites
• Not run Cloud P0 locally: no Doris regression cluster available in this workspace

[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]

run buildall

[shuke987]

/review

[github-actions]

Reviewed PR #65208.

I found no blocking issues and no inline comments to submit.

Critical checkpoint conclusions:

• Goal and tests: the PR batches the generated INSERT ... VALUES setup rows in the three changed export_p0 suites while preserving the same row sets. Existing ordered qt_select_default checks and OUTFILE comparisons still validate the exported results, and no .out update is needed because expected query output is unchanged.
• Scope: the change is small and limited to regression-test setup code in test_outfile_native.groovy, test_outfile.groovy, and test_outfile_expr.groovy.
• Concurrency/lifecycle/config/compatibility/persistence: no product runtime code, persisted format, FE/BE protocol, or configuration item is changed. The added batching executes sequentially in the same suite setup path.
• Parallel or related paths: checked the changed local/native export suites and analogous outfile test patterns; I did not find a distinct parallel-path issue introduced by this PR.
• Test results and style: git diff --check is clean for the base/head diff. I did not run the regression suite because this checkout lacks thirdparty/installed, thirdparty/installed/bin/protoc, and a Doris regression cluster.
• Performance: replacing one large literal VALUES statement with 500-row batches addresses the stated Cloud P0 planning timeout risk without changing the tested output semantics.

Subagent conclusions:

• optimizer-rewrite: no optimizer/rewrite correctness, semantic equivalence, or parallel join/aggregate issue found; final convergence returned NO_NEW_VALUABLE_FINDINGS.
• tests-session-config: no regression expected-output, session/config propagation, compatibility, or CI/style issue found; final convergence returned NO_NEW_VALUABLE_FINDINGS.

User focus points: no additional user-provided review focus was present.

[HappenLee]

LGTM