feat: Add retroactive trace linking for orphaned spans

[harry-rhesis]

Overview

Add a background job to link orphaned spans that are missing test_result_id as a safety net for edge cases where force flush fails.

Problem

While force flush solves the primary race condition, edge cases can still occur:

• Network failures during span export
• SDK crashes before flush completes
• Timeout scenarios
• Backend downtime during export

Proposed Solution

Implement a periodic Celery task that:

1. Finds spans with test context but null test_result_id (created in last hour)
2. Groups them by (test_run_id, test_id, test_configuration_id, org_id)
3. Looks up matching test results from database
4. Links the orphaned spans using update_traces_with_test_result_id()

Implementation Outline

File: apps/backend/src/rhesis/backend/tasks/telemetry/retroactive_linking.py

@celery_app.task
def link_orphaned_traces():
    # Find spans: test_id IS NOT NULL AND test_result_id IS NULL
    # Created within last hour
    # Group by execution context
    # For each group, find test_result and update spans
    pass

Schedule to run every 1-5 minutes via Celery beat.

Benefits

• Eventual consistency guarantee
• Handles all edge cases
• Zero impact on test execution latency
• Complements force flush approach

Priority

Low - Force flush handles 99.9% of cases. This is for remaining edge cases.