[BUG] Cluster dead because detector shard stuck initializing

What is the bug?
Upgraded a cluster from 2.11.1 to 2.15.0 and the cluster is in red status because the .opensearch-sap-network-detectors-queries-000007 shard is stuck initializing.

{
    'current_node': {
        'attributes': {'shard_indexing_pressure_enabled': 'true'},
        'id': 'zLNHMkpeReSxfbVn99y1Sw',
        'name': 'opensearch-node1',
        'transport_address': '172.19.0.3:9300',
    },
    'current_state': 'initializing',
    'explanation': 'the shard is in the process of initializing on node [opensearch-node1], wait until initialization has completed',
    'index': '.opensearch-sap-network-detectors-queries-000007',
    'primary': True,
    'shard': 0,
    'unassigned_info': {
        'at': '2024-07-24T12:00:27.551Z',
        'last_allocation_status': 'throttled',
        'reason': 'CLUSTER_RECOVERED',
    },
}

How can one reproduce the bug?
Steps to reproduce the behavior:
No idea. Create cluster at 2.11, add detection rules, upgrade to 2.15 and observe error.

What is the expected behavior?
My cluster to not die because a stupid feature can't start.
If anomaly detection is broken, then only let that feature be broken, not my entire cluster!!!!!! FFS
Separate user indices from system indices. The fact that this isn't done and that they are treated identical is a super stupid decision.

What is your host/environment?

• Docker container for 2.15

Do you have any screenshots?
If applicable, add screenshots to help explain your problem.

Do you have any additional context?
Add any other context about the problem.

Start by investigating the reported upgrade path from 2.11.1 to 2.15.0 with detection rules and inspect the shard-allocation state for .opensearch-sap-network-detectors-queries-000007. Confirm why the detector shard remains initializing and define completion as the cluster staying available even when this feature cannot start.