fix: synchronize CoalesceService slot lifecycle to prevent dropped reruns

[JuliusHenke]

Motivation

• A recent refactor introduced a race in CoalesceService where submissions near task completion could create an orphan nextFuture (never completed) or allow duplicate concurrent executions for the same key.
• This affects cache rebuilding and relay-detail recomputation jobs and can cause stale caches, hung CompletableFuture.allOf callers, duplicate DB/cache work, and extra lock contention.

Description

• Serialize submitAsync state transitions (running, rerunRequested, nextFuture) by synchronizing on slot.monitor so the initial-run path sets running and prepares the first future while holding the lock.
• Start the async run loop only after the monitor-held state is established to avoid submitters observing partially-shutdown state.
• Move the worker-loop rerun check and shutdown (running=false + slots.remove) into the same synchronized(slot.monitor) block and promote/clear nextFuture there to prevent orphaned futures and stale-slot removal races.
• Change is limited to backend/src/main/kotlin/org/tormap/service/CoalesceService.kt and preserves the intended coalescing semantics.

Testing

• Attempted ./gradlew :backend:test --tests "org.tormap.service.CoalesceServiceTest" at the repo root failed because ./gradlew was not present in that path. (failed)
• Ran cd backend && ./gradlew test --tests "org.tormap.service.CoalesceServiceTest" which failed due to a Gradle/toolchain environment error (25.0.2), so unit tests could not be executed in this environment. (failed)

---

Codex Task

[Copilot]

Pull request overview

This PR adjusts the concurrency control in CoalesceService to better synchronize per-key slot lifecycle and prevent rerun futures from being dropped or orphaned during task completion races.

Changes:

• Synchronizes submitAsync state transitions on slot.monitor before starting the worker loop.
• Moves rerun promotion and slot shutdown/removal into a monitor-held critical section in the worker loop.
• Refactors the “first run” path to create/return its CompletableFuture only after the slot state is established.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Just as a heads up, I was blocked by some firewall rules while working on your feedback. Expand below for details.

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• 113.114.104.172.in-addr.arpa
  • Triggering command: /usr/lib/jvm/temurin-17-jdk-amd64/bin/java /usr/lib/jvm/temurin-17-jdk-amd64/bin/java -Dorg.gradle.internal.worker.tmpdir=/home/REDACTED/work/tormap/tormap/backend/build/tmp/test/work @/home/REDACTED/.gradle/.tmp/gradle-worker-classpath4577654228264605735txt -Xmx512m -Dfile.encoding=UTF-8 -Duser.country -Duser.language=en -Duser.variant -ea worker.org.gradle.process.internal.worker.GradleWorkerMain 'Gradle Test Executor 2' (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

[JuliusHenke]

@copilot perform a final review to determine if there are any issues left

[Copilot]

@copilot perform a final review to determine if there are any issues left

Final review done at 02ebc1c.

Current state:

• The earlier CI compile failure on an older commit was due to continue in an inline lambda and is already resolved in the latest commit.
• Latest backend workflow run for this head SHA is still queued (actions/runs/26177660781), so no new CI result yet.
• One issue still appears intermittently in focused local runs: CoalesceServiceTest > never runs the same key concurrently even under repeated submissions can fail (expected 2, got 1).

So I still see that intermittent test-stability concern as the remaining issue.