fix: release query resources and drain Store RPCs - #3294

- keep caller transactions open when traversers close
- rollback failed worker transactions and reset thread state
- run Store shutdown tests through the active Maven suite
- cover task commit and cleanup with real RocksDB tests

- require the central server profile and test report
- verify every selected Store profile has a session
- retain strict execution order and coverage checks

- reject interrupted scans without partial success
- wake closed channels without interrupting RPC callers
- verify cancellation status through active scan paths
- retain library and executable Node artifacts

- expose stop diagnostics from smoke cleanup
- preserve failure status and retained PID evidence
- retain existing native shutdown timeout

- include the merged runtime alignment
- retain lifecycle changes as a separate PR
- inherit current upstream CI configuration
- preserve both upgrade and lifecycle documentation

- keep the ordinary code margin at 120 columns
- allow Markdown paragraphs without hard wrapping
- show a 160-column Markdown visual guide
- align CodeRabbit instructions with repository style

- separate response completion from resource release
- isolate cancellation callback failures
- preserve native iterator close failures
- transfer iterator ownership before prefetch
- cover Spring shutdown and real JNI boundaries

- Keep the newest run for each workflow and PR
- Verify source freshness before automatic retries
- Test retry decisions and document the CI policy

- Keep one active run per workflow and PR
- Preserve push and manual workflow executions
- Cover CI workflows on this PR target branch

- Reserve the exec classifier for distribution assembly
- Avoid compile-time resolution of packaged artifacts
- Keep the existing assembly filename and runtime inclusion

- separate natural completion from cancellation
- retain failed cleanup owners until retry succeeds
- close nested scan iterators and serialize receipts
- verify cancellation and ordered scan backpressure

- deliver credit-backed results after request half-close
- cancel client streams when iterators close early
- preserve queued final results across completion races
- verify matched client and server stream behavior

- report outstanding scan count at the wait interval
- retain details only for actual cleanup failures
- preserve shutdown waits and interrupt handling

- preserve cancellation precedence in terminal callbacks
- finish cancelled scans after automatic iterator cleanup
- cover blocked close with callback and unary gRPC tests
- document cancellation and retain native drain ownership

- capture the observed wait state before asserting
- retain database and failed-cleanup ownership checks
- verify scan and aggregate shutdown regressions

- adopt the Java 17 baseline
- retain Gremlin and lifecycle semantics
- preserve both test-suite additions

- propagate transport cancellation independently
- reject repeated queries before iterator allocation
- cover real cancellation and single terminal callbacks
- document request and response lifecycle boundaries

- wait until response callback is in flight
- trigger cleanup failure in the intended order
- preserve worker ownership and terminal assertions

- close write batch and options at final detach
- retain owners while another lease is active
- guarantee both disposal attempts and closed state
- cover real JNI owners across repeated worker requests

- select supported auth fixtures by actual graph backend
- run cleanup assertions without a backend system property
- give the OLAP sentinel fixture a valid vertex ID
- assert the intended schema rejection source

- retain schema identity across request cleanup
- keep cache invalidation listeners until graph close
- dispose only the matching graph listener generation
- cover schema append and API update regressions

- observe normal sender completion before teardown
- avoid inline transport cancellation during completion
- retain early cancel and resource release assertions

- check published terminal state before timed poll
- preserve final-batch queue visibility ordering
- cover exhausted single and merged streams
- register the regression in the client suite

- detach owned store listeners without a cache holder
- preserve reopened listener generation ownership
- clear auth projects using the actual graph backend
- retain reopen failures and clean temporary data

- propagate transaction cleanup failures after all owners close
- retain graph-scoped schema cache listeners
- cancel all started result streams on close
- register graph scans for deterministic shutdown cleanup

- retain merged Topling and shared schema behavior
- drain Store streams before storage shutdown
- preserve cleanup failures and update regressions

- exclude HTTP libraries from Node test dependency
- preserve production dependencies and security gate
- verify Store lifecycle tests without web jars