[13.x] Optionally flush the SQS overflow store on queue:clear

[Orrison]

Right now, queue:clear purges SQS but leaves the offloaded payloads sitting in the cache store. Even if a TTL got added later, when someone clears the queue, they may want the storage reclaimed now rather than waiting for expiration (matters especially on S3, where leftover objects keep costing money or persistence of payloads with sensitive data could be an issue).

Adds an opt-in flush_on_clear flag under the overflow config. When on, SqsQueue::clear() calls flush() on the configured cache store after purging SQS. Defaults to off.

Worth flagging: for most cache stores flush() wipes the whole store, not just the overflow entries. I considered some options to narrow its scope. Like tags or maybe a manifest file/entry pattern. Tags didn't seem right (though maybe that is still an option), and a manifest adds per-push overhead and concurrency issues that felt heavier than the feature needs. So this is opt-in and separate from delete_after_processing, with the recommendation to point overflow.store at a dedicated cache store (the new storage driver with its own disk and/or path is a clean fit).

Tests cover flush runs when enabled, don't run when disabled or when overflow is off, and forward the configured store name.

Has any user-facing documentation been written for the overflow feature yet? Happy to put together a docs PR if not.

[github-actions]

Thanks for submitting a PR!

Note that draft PRs are not reviewed. If you would like a review, please mark your pull request as ready for review in the GitHub user interface.

Pull requests that are abandoned in draft may be closed due to inactivity.