fix: rate limit unsupported source issue creation

[kentoku24]

Motivation

• The /add flow invoked GitHub issue creation on every unsupported-source error with no local gating, enabling spam and GitHub API token/rate exhaustion.
• Introduce a minimal in-process guard to reduce abuse while preserving existing user-visible behavior and reporter semantics.

Description

• Add a host-based cooldown to AddCommandHandler with a default of DEFAULT_UNSUPPORTED_SOURCE_REPORT_COOLDOWN_SECONDS = 3600 and a configurable unsupported_source_report_cooldown_seconds property.
• Implement _last_reported_hosts tracking and a now injection point so the cooldown check uses now() and is testable.
• Short-circuit calls to unsupported_source_reporter.report_unsupported_source when the host was reported recently and return a user-facing rate-limit message via UNSUPPORTED_SOURCE_REPORT_RATE_LIMIT_MESSAGE while keeping existing error/result messages intact.
• Add a deterministic unit test test_start_rate_limits_unsupported_source_issue_reporting_per_host to validate throttling and update imports/constants used by tests.

Testing

• Ran python -m unittest tests.test_discord_add tests.test_discord_interactions tests.test_github_issue_reporting and all tests passed (36 tests ran and returned OK).
• New unit test exercises time-injected paths to assert the reporter is called only once across rapid repeated unsupported-source submissions for the same host.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2d5005d20b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Guard host cooldown map with a lock

_report_unsupported_source performs a read-then-write on _last_reported_hosts without synchronization, so concurrent requests can both pass the cooldown check and both call report_unsupported_source for the same host. In this repo the service runs on ThreadingHTTPServer (manga_watch/run_service.py), so this race is reachable in production and undermines the new rate-limit goal (extra GitHub API calls and possible duplicate issue creation under burst traffic). Please make the check/update atomic (e.g., lock per handler or per-host reservation) so the throttle is effective across threads.

Useful? React with 👍 / 👎.