[cuebot] Add scheduled subscription recalculation task

[pmady]

Summary

This PR adds a scheduled task to periodically call the recalculate_subs() database function, eliminating the need for an external cronjob on the database server.

Problem

Subscription accountability has a condition where, when running at large scale, values can become out of date. The recalculate_subs() function was added in PR #1380 to fix this, but it needs to be called on a schedule. Currently, this requires an external cronjob on the database server, which is not ideal.

Solution

This PR integrates the subscription recalculation into cuebot's existing Quartz scheduler infrastructure:

1. MaintenanceTask enum - Added LOCK_SUBSCRIPTION_RECALCULATION for database-level locking to prevent concurrent execution across multiple cuebot instances
2. MaintenanceDao - Added recalculateSubscriptions() method to call the PostgreSQL function
3. MaintenanceManagerSupport - Added recalculateSubscriptions() method with proper locking
4. Quartz configuration - Added job and trigger to run every 2 hours (configurable)
5. Configuration property - Added maintenance.subscription_recalculation_interval_ms (default: 2 hours)
6. Database migration - Added V35 migration to insert the task_lock entry

Configuration

The interval can be configured via the maintenance.subscription_recalculation_interval_ms property in opencue.properties. Default is 7200000 ms (2 hours), matching SPI's production setup.

Multi-instance Support

The implementation uses the existing task_lock mechanism to ensure only one cuebot instance runs the recalculation at a time, even in multi-instance deployments.

Closes #1562

[pmady]

CI Failure Analysis

The failing checks (Build Cuebot and Run Unit Tests and Run Integration Test) are due to transient infrastructure issues with the Spring repository, not related to the code changes in this PR.

Error Details

The build is failing with HTTP 401 errors when trying to access repo.spring.io/plugins-snapshot:

Failed to get resource: GET. [HTTP HTTP/1.1 401 : https://repo.spring.io/plugins-snapshot/org/eclipse/ee4j/project/1.0.5/project-1.0.5.pom)]
Failed to get resource: GET. [HTTP HTTP/1.1 401 : https://repo.spring.io/plugins-snapshot/org/springframework/spring-framework-bom/5.2.1.RELEASE/spring-framework-bom-5.2.1.RELEASE.pom)]

Verification

The latest master branch run (#20526703128) passed successfully, confirming this is a transient issue.

Request

Could a maintainer please re-run the failed jobs when the Spring repository is accessible again? Thank you!

[lithorus]

Looks like it fails the spotless java checks.

[DiegoTavares]

Just a heads up, this PR is still failing CICD with the following error. I'm moving it to Draft state. Please only remove the draft state when it is ready for review.

java.lang.IllegalArgumentException: Could not resolve placeholder 'maintenance.subscription_recalculation_interval_ms' in value "${maintenance.subscription_recalculation_interval_ms}"
                at org.springframework.util.PropertyPlaceholderHelper.parseStringValue(PropertyPlaceholderHelper.java:178)
                at org.springframework.util.PropertyPlaceholderHelper.replacePlaceholders(PropertyPlaceholderHelper.java:124)
                at org.springframework.beans.factory.config.PropertyPlaceholderConfigurer$PlaceholderResolvingStringValueResolver.resolveStringValue(PropertyPlaceholderConfigurer.java:230)
                at org.springframework.beans.factory.config.BeanDefinitionVisitor.resolveStringValue(BeanDefinitionVisitor.java:296)
                at org.springframework.beans.factory.config.BeanDefinitionVisitor.resolveValue(BeanDefinitionVisitor.java:217)
                at org.springframework.beans.factory.config.BeanDefinitionVisitor.visitPropertyValues(BeanDefinitionVisitor.java:147)
                at org.springframework.beans.factory.config.BeanDefinitionVisitor.visitBeanDefinition(BeanDefinitionVisitor.java:85)
                at org.springframework.beans.factory.config.PlaceholderConfigurerSupport.doProcessProperties(PlaceholderConfigurerSupport.java:225)
                ... 51 more

[pmady]

@DiegoTavares fixed the missing parameter, now CI looks good. Thanks

[DiegoTavares]

@htehara and @ramonfigueiredo FYI

[DiegoTavares]

@htehara can you review the migrations to confirm a task_lock is the best mechanism to prevent multiple executions?