Cluster Index Replication health check doesn't work for a database in a different timezone than the nodes

Issue Summary

Cluster Index Replication health check reports false-positives when cluster and database are in different timezones.

Steps to Reproduce

1. Configure a Jira cluster in GMT+0 servers
2. Configure a database in a GMT-5 server
3. Run the instances and see the health check reports a 18,000 seconds delay in Index Replication

Expected Results

The health check would compare nodes and db dates converting them to the same TZ.

Actual Results

The health check compares the nodes current time (from the JVM) with the replicatedindexoperation.index_time from the database, which is stored in database local time.
Thus the health check will always report a delay in the same amount of seconds as the timezone offset.

Workaround

No current workaround aside from configuring the database and nodes to the same timezone or ignoring the alarms if they fall within the timezone offset threshold (eg 18,000 and something seconds for a 5h difference). 

More details in the KB article: Cluster Index Replication health check reports delay when the database is in a different timezone than Jira

Notes

The problem was introduced as part of DBR changes in 8.12.