fix: O3-5485 Fix NaN return from computeAverageWaitTimeInMinutes when no entries have endedAt

[sanks011]

Summary

QueueUtils.computeAverageWaitTimeInMinutes() returns Double.NaN when all queue entries
in the list have a null endedAt date. This causes the /ws/rest/v1/queue-entry-metric and
/ws/rest/v1/queue-metrics REST endpoints to return NaN in JSON responses - which is
invalid per RFC 7159 and breaks frontend metric display.

Root Cause

The method computes totalWaitTime / numEntries without guarding against numEntries == 0.
When every entry is skipped (because endedAt is null), this evaluates to 0.0 / 0 which
per JLS §15.17.2 produces Double.NaN — no exception is thrown.

Fix

Added a numEntries > 0 guard before the division. When no entries have both startedAt
and endedAt populated, the method now returns 0.0 instead of NaN.

Tests Added

• computeAverageWaitTime_shouldReturnZeroForNullList
• computeAverageWaitTime_shouldReturnZeroForEmptyList
• computeAverageWaitTime_shouldReturnZeroWhenAllEntriesHaveNullEndedAt
• computeAverageWaitTime_shouldReturnCorrectAverageForEntriesWithEndedAt
• computeAverageWaitTime_shouldSkipEntriesWithNullEndedAt

##Related
https://openmrs.atlassian.net/jira/software/c/projects/O3/boards/37?selectedIssue=O3-5485

[vimla01]

I have Tested this locally and this looks correct. It fixes the NaN case by returning 0.0 when no matching entries have endedAt, and the added tests cover the expected edge cases.

small note : a shorter branch name can make manual fetch/checkouts a bit easier.

[sanks011]

I have Tested this locally and this looks correct. It fixes the NaN case by returning 0.0 when no matching entries have endedAt, and the added tests cover the expected edge cases.

small note : a shorter branch name can make manual fetch/checkouts a bit easier.

Thanks for testing and the feedback! Good tip on the branch name - I'll keep it shorter next time around. 👍