[FLINK-40368][table-planner] Fix deduplication changelog inference

[ltyfulan9]

What is the purpose of the change

Streaming deduplication represented by a Top-1 Rank can be incorrectly inferred as INSERT_ONLY when its ORDER BY uses a non-time attribute or multiple columns. This can cause downstream operators to miss required updates and retractions. This change restricts INSERT_ONLY inference to the existing single-time-attribute case.

Brief change log

• Gate deduplication INSERT_ONLY changelog inference with RankUtil.sortOnTimeAttributeOnly.
• Preserve ALL_CHANGES for non-time-attribute and multi-column ORDER BY keys.

Verifying this change

• Existing plan regressions cover RankTest#testDeduplicateOnNonTimeAttributeGeneratesUpdates and RankTest#testDeduplicateOnMultipleColumnsGeneratesUpdates.
• flink-table-planner Spotless check passes.
• Full reactor compilation was not completed locally because the configured Maven repository could not resolve flink-core:tests:2.4-SNAPSHOT before reaching Table Planner.

Does this pull request potentially affect one of the following parts:

• Dependencies: no
• Public API: no
• Serializers: no
• Runtime per-record code paths: no
• Deployment or recovery: no
• S3 file system connector: no

Documentation

• New feature: no
• Documentation: not applicable

---

Was generative AI tooling used to co-author this PR?

• Yes (Codex GPT-5)

Generated-by: Codex GPT-5

[flinkbot]

CI report:

• 92f6f05 Azure: SUCCESS

Bot commands

The @flinkbot bot supports the following commands:

• @flinkbot run azure re-run the last Azure build

[ltyfulan9]

@flinkbot run azure