[NEEDS CODE REVIEWER] fix: Handle failed status requests in SABnzbd check_connected

[lolimmlost]

Summary

Mirrors the qBit fix from #342 (just approved) for the same bug class in SABnzbd.

Previously, an HTTP error from SABnzbd's status endpoint (e.g. a 500) would raise out of make_request and crash decluttarr entirely. Now the error is caught and treated as a disconnected client, so the cycle is skipped and retried next run.

Before: SAB returns 500 → HTTPError propagates up through check_connected → _check_client_connection_status → _download_clients_connected → removal_jobs → run_jobs → main → process exits, container restart-loops.

After: SAB returns 500 → caught locally → warning logged → False returned → that client is skipped this cycle, other clients/instances continue, next cycle retries.

Same shape as #342 — granular per-client handling, not a broad top-level catch.

Test plan

Smoke tested four paths against the patched check_connected:

• HTTPError (500) → returns False, warning logged, no crash
• Connected ({"status": True}) → returns True
• Missing status key → returns False
• Invalid JSON (response.json() raises) → returns False

Full test suite: 198/198 passing.

Notes

• No new tests added — there's no existing coverage for check_connected in either qbit or sabnzbd, so this stays consistent with the codebase. Happy to add tests for both if preferred.
• Targets dev for consistency with [NEEDS CODE REVIEWER] Handle failed /sync/maindata requests in check_connected #342.

🤖 Generated with Claude Code

[ManiMatter]

@jrhager84 would you be ok to review this?

[jrhager84]

Yeah - It's on the docket for this week.