[FLINK-40111][table] Fall back to retract for a keyless sink without an upsert key

[gustavodemorais]

What is the purpose of the change

A sink without a primary key that advertises an upsert changelog (ONLY_UPDATE_AFTER) can only apply UPDATE_AFTER when the input carries an upsert key for the downstream to fold by. When the input has no upsert key, the planner now falls back to BEFORE_AND_AFTER so the query plans as retract (when the input can produce UPDATE_BEFORE) instead of emitting an unkeyed upsert stream. Keyed-sink behavior is unchanged.

Brief change log

• Fall back to retract for a keyless sink whose input has no upsert key in canUpsertKeysWithImmutableColsSatisfyPk.
• Switch the keyless-union DagOptimizationTest case to a retract sink and regenerate its plan.

Verifying this change

• ChangelogModeInferenceTest
• DagOptimizationTest

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

• Dependencies (does it add or upgrade a dependency): (no)
• The public API, i.e., is any changed class annotated with @Public(Evolving): (no)
• The serializers: (no)
• The runtime per-record code paths (performance sensitive): (no)
• Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: (no)
• The S3 file system connector: (no)

Documentation

• Does this pull request introduce a new feature? (no)
• If yes, how is the feature documented? (not applicable)

---

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

• Yes (please specify the tool below)

2.1.196 (Claude Code) with Opus 4.8

[flinkbot]

CI report:

• 23bbe0f Azure: SUCCESS

Bot commands

The @flinkbot bot supports the following commands:

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

[twalthr]

The change looks good to me. This edge case behavior was undefined before and by looking at the test also untested. The change is a good preparation for developing a more dynamic collect() sink in the future that does not dictate retract but dynamically adjusts to what the upstream pipeline actually demands.