Add HTTP-based cron monitoring support

[notheotherben]

Summary

Adds support for reporting the lifecycle of each scheduled backup run to an HTTP-based cron monitoring service (such as Sentry Cron Monitors or healthchecks.io), as requested.

The application makes a simple HTTP GET request to a configured endpoint whenever a scheduled run starts and again when it completes — reporting success if every policy ran cleanly and failure if any policy produced one or more errors.

Configuration

A new top-level ping key accepts an optional URL for each state. Any state you leave out is simply not reported:

schedule: "0 * * * *"

ping:
  start: https://sentry.io/api/0/organizations/your-org/monitors/github-backup/checkins/?status=in_progress
  success: https://sentry.io/api/0/organizations/your-org/monitors/github-backup/checkins/?status=ok
  failure: https://sentry.io/api/0/organizations/your-org/monitors/github-backup/checkins/?status=error

backups:
  - kind: github/repo
    from: user
    to: /backup/github
    credentials: !Token your_access_token

Allowing a separate URL per state keeps the implementation to plain GET requests while supporting both query-parameter-based services (Sentry: ?status=...) and path-suffix-based services (healthchecks.io: /start, /fail).

Behaviour notes

• Best-effort. If the monitoring service can't be reached (or returns a non-2xx status), a warning is logged but the backup run itself is never affected. A flaky monitor can't cause an otherwise healthy run to be reported as failed.
• Interrupted runs. If a run is interrupted via SIGINT, neither success nor failure is reported (the monitor will naturally observe a missed check-in).
• A run is reported as failure when any policy reports one or more errors, tracked via the pairing handler's error count.

Implementation

• New src/ping.rs module with PingConfig (deserialized from the ping key) and a Pinger that issues the GET requests.
• Config gains a ping field.
• main.rs instantiates the pinger and reports start before each run and success/failure afterwards; LoggingPairingHandler now accumulates an error count to drive that decision.

Testing

• Unit tests for PingConfig deserialization.
• wiremock-backed tests verifying each state hits the right URL, that unconfigured states make no request, and that error responses are swallowed without panicking.
• Config tests covering the new ping section (present and absent).
• A unit test for LoggingPairingHandler's error counting (the success/failure decision logic).

cargo clippy and cargo test --features pure_tests both pass cleanly. (The repository's network-dependent GitHub API tests fail in the sandbox due to no outbound access, unrelated to this change; they are gated by the pure_tests feature.)

Docs

• New "Cron Monitoring" section in the README.
• New docs/guide/monitors.md guide page (added to the VuePress sidebar) with Sentry and healthchecks.io examples.
• Example added to examples/config.yaml.

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 92.17391% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 72.74%. Comparing base (43f0cb0) to head (2f87213).

Files with missing lines	Patch %	Lines
src/main.rs	62.50%	9 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #597      +/-   ##
==========================================
+ Coverage   70.87%   72.74%   +1.87%     
==========================================
  Files          28       29       +1     
  Lines        1971     2077     +106     
==========================================
+ Hits         1397     1511     +114     
+ Misses        574      566       -8     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.