[BUG] Alerts break and alerting lock index becomes unrecoverable

[rkbennett]

Describe the bug

When using alerting the .opensearch-alerting-config-lock shard isn't rotated and can grow past a recoverable shard size which then breaks all alerting. Because this is a shard that isn't managed by the user, a rotation strategy should be default

Related component

Storage

To Reproduce

1. Create an alert
2. Restart the cluster once the shard size has grown past the limit_size_in_bytes setting
3. .opensearch-alerting-config-lock shard will never recover and alerts will no longer work

Expected behavior

System shards shouldn't get into permanently broken states

Additional Details

Plugins
Security

Host/Environment (please complete the following information):

• OS: Opensearch 2.18 container image
• Version 2.18

[dblock]

[Catch All Triage - 1, 2, 3]