[Fix][connector-fake] Signal no more splits to avoid FakeSourceReader wait-split hang after restore

[yzeng1618]

Purpose of this pull request

Fix a potential hang after Flink restart/restore when using FakeSource: once split discovery is finished, some readers may receive no splits (e.g. all splits were already assigned before restore). Previously, FakeSourceSplitEnumerator only called signalNoMoreSplits when it actually assigned splits, which could leave readers waiting forever.

Does this PR introduce any user-facing change?

Yes (bug fix). Previously jobs could hang after restart/restore; now readers can finish correctly.

How was this patch tested?

Added unit tests: FakeSourceSplitEnumeratorTest.java

Check list

• [*] If any new Jar binary package adding in your PR, please add License Notice according
  New License Guide
• If necessary, please update the documentation to describe the new feature. https://github.com/apache/seatunnel/tree/dev/docs
• If necessary, please update incompatible-changes.md to describe the incompatibility caused by this PR.
• If you are contributing the connector code, please check that the following files are updated:
  1. Update plugin-mapping.properties and add new connector information in it
  2. Update the pom file of seatunnel-dist
  3. Add ci label in label-scope-conf
  4. Add e2e testcase in seatunnel-e2e
  5. Update connector plugin_config

[dybyte]

This seems duplicate with #10208 .

[yzeng1618]

This seems duplicate with #10208 .

Thanks for the heads-up! I checked #10208.

#10208 adds a framework-level re-signal: if a reader re-registers after failover and NoMoreSplitsEvent was already signaled before, the engine/flink translation will re-signal it to that reader.

This PR (#10275) fixes a different issue in FakeSourceSplitEnumerator: it previously called signalNoMoreSplits() only when it actually assigned splits. After restore (or late registration), a reader may receive zero splits, so it never gets the end-of-input signal and can wait forever (see the added unit test signalNoMoreSplitsAfterRestoreWhenNoPendingSplits).

So they’re complementary: #10275 ensures FakeSource always produces signalNoMoreSplits for all registered readers after discovery (even when no split is assigned), while #10208 ensures previously signaled NoMoreSplits can be re-delivered after failover.