[To dev/1.3] [Pipe] Fix processor worker starvation on pipe stop (#18396)

[Caideyipi]

Description

Cherry-pick 3c0c7fef0068162d7b3fc582df61a82050382ac8 from master to dev/1.3.

Original PR: #18396

• Add an epoch-based execution guard so a processor invocation that observes STOP yields even if the Pipe is immediately started again.
• Check the guard around TsFile-close waits, parser admission, tablet boundaries, and OOM backoff while preserving the current event and parsing progress.
• Make parser admission non-blocking for processor subtasks so one waiting Pipe does not starve another Pipe sharing the same worker.
• Cancel pending parser requests on STOP and release active parser reservations before yielding.
• Watch processor workers for continuously running events and emit bounded, rate-limited stack diagnostics.

Backport adaptations

• Wire the scheduled worker watcher through the dev/1.3 subtask executor API because that scheduler support predates this change on master.
• Retain the dev/1.3 inline logging style instead of pulling in the newer i18n message layer.
• Adapt the new tests to the dev/1.3 Pipe pattern, TsFile, and Java 8 APIs.

Verification

• Spotless: passed for iotdb-core/node-commons and iotdb-core/datanode.
• Java 8 main/test compilation: passed for both affected modules; Checkstyle reported 0 violations.
• Focused tests: 23 passed, 0 failures:
  • PipeProcessorSubtaskWorkerTest
  • PipeProcessorSubtaskExecutionGuardTest
  • PipeTsFileInsertionEventAdmissionTest
  • PipeMemoryManagerTest
  • PipeProcessorSubtaskExecutorTest

---

This PR has:

• been self-reviewed.
• added or updated unit tests.