Limit media download retries to prevent unbounded background rescheduling

[Szpadel]

Motivation

• The existing retry flow scheduled detached retry tasks indefinitely when a download failed and media_path remained None, enabling unbounded background work and a potential DoS.
• A minimal, localized fix is needed to cap automatic retries while preserving normal enqueue behavior and backwards compatibility.

Description

• Add a MAX_RETRY_ATTEMPTS: u8 = 3 constant and a retry_attempt: u8 field to FetchMediaWorkerArgs with #[serde(default)] so older queued jobs default to 0 when deserialized. (src/workers/fetch_media.rs)
• Change schedule_media_retry to accept the current retry_attempt, skip scheduling when the cap is reached, and re-enqueue the job with retry_attempt incremented. (src/workers/fetch_media.rs)
• Initialize retry_attempt: 0 at explicit enqueue call sites to maintain behavior for new jobs. (src/workers/fetch_source_info.rs, src/controllers/media.rs)
• Add unit tests that assert retry_attempt defaults to 0 on deserialization and that MAX_RETRY_ATTEMPTS is positive. (src/workers/fetch_media.rs tests)

Testing

• Ran code formatting with cargo fmt --all which completed successfully.
• Ran cargo test fetch_media_worker_args_retry_attempt_defaults_to_zero and it passed.
• Ran cargo test retry_attempt_limit_is_greater_than_zero and it passed.
• Ran existing retry scheduler tests such as cargo test executes_action_when_check_passes and they passed, confirming no regression in RetryScheduler behavior.

---

Codex Task