[fix](be) Avoid finalized pipeline task submit crash

[BiteTheDDDDt]

What problem does this PR solve?

Issue Number: None

Related PR: #64899

Problem Summary: Backport #64899 to branch-4.0. A runtime filter dependency can submit a pipeline task after the task has started close/finalize. On branch-4.0, HybridTaskScheduler::submit() calls PipelineTask::is_blockable(), which reads task sink/operator resources that finalize releases. This adds a submit gate so terminal close/finalize rejects late hybrid scheduler submits before is_blockable() can touch released resources.

Release note

None

Check List (For Author)

• Test: Unit Test / Static check
  • build-support/clang-format.sh be/src/pipeline/pipeline_task.cpp be/src/pipeline/pipeline_task.h be/src/pipeline/task_scheduler.cpp be/test/pipeline/pipeline_task_test.cpp && build-support/check-format.sh be/src/pipeline/pipeline_task.cpp be/src/pipeline/pipeline_task.h be/src/pipeline/task_scheduler.cpp be/test/pipeline/pipeline_task_test.cpp
  • ./run-be-ut.sh --run --filter=PipelineTaskTest.TEST_CLOSED_TASK_REJECTS_HYBRID_SUBMIT_BEFORE_FINALIZE:PipelineTaskTest.TEST_FINALIZED_TASK_REJECTS_HYBRID_SUBMIT
• Behavior changed: No
• Does this need documentation: No

[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?

[BiteTheDDDDt]

/review

[BiteTheDDDDt]

run buildall

[github-actions]

I reviewed PR #64946 across the four changed files and the surrounding pipeline lifecycle paths. I did not find a blocking correctness issue or any inline comment worth submitting.

Critical checkpoints:

• HybridTaskScheduler::submit() now checks the submit gate under _blockable_check_lock before calling PipelineTask::is_blockable(), so late submits after terminal close/finalize no longer read _sink or _operators.
• PipelineTask::close(close_sink=true) and PipelineTask::finalize() close the submit gate before terminal resource release, while close(close_sink=false) keeps normal EOS/pending-finish resubmits available.
• The new tests cover both close-before-finalize and post-finalize submit paths through the real HybridTaskScheduler::submit() call, including a counting sink assertion that is_blockable() is not evaluated after the gate closes.
• Existing review context for this PR had no inline threads, and there was no additional user focus beyond the full PR review.

Validation:

• git diff --check 5494d4eca3fc5ee0f19f3371a16ed21430c7bf83..314d0c451fc97a6771925d4b4d8ff1a86909258f -- be/src/pipeline/pipeline_task.cpp be/src/pipeline/pipeline_task.h be/src/pipeline/task_scheduler.cpp be/test/pipeline/pipeline_task_test.cpp passed.
• I could not run the BE unit tests in this runner because thirdparty/installed, thirdparty/installed/bin/protoc, and thirdparty/installed/bin/thrift are missing.
• I did not rely on local format tooling because this runner has clang-format 18.1.3 while the Doris scripts normally expect clang-format 16.

Subagent conclusions:

• optimizer-rewrite: no optimizer/rewrite, semantic-equivalence, or parallel join/aggregate issue found; convergence round 1 returned NO_NEW_VALUABLE_FINDINGS.
• tests-session-config: no test/session/config/style issue found; convergence round 1 returned NO_NEW_VALUABLE_FINDINGS.

[hello-stephen]

BE Regression && UT Coverage Report

Increment line coverage 100.00% (17/17) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	71.58% (25594/35758)
Line Coverage	54.45% (271492/498633)
Region Coverage	51.97% (224871/432716)
Branch Coverage	53.45% (96912/181319)

[hello-stephen]

BE Regression && UT Coverage Report

Increment line coverage 100.00% (17/17) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	71.57% (25593/35758)
Line Coverage	54.45% (271507/498633)
Region Coverage	51.98% (224911/432716)
Branch Coverage	53.45% (96917/181319)