Notifications were fire-and-forget: if Slack, Mattermost, or ntfy was
temporarily down, changes were silently lost. This adds automatic retry
with exponential backoff and jitter to all notification endpoints.
Implementation:
- New retry.go with configurable RetryConfig (max retries, base delay,
max delay) and exponential backoff with ±25% jitter
- Each dispatch goroutine now wraps its send call in deliverWithRetry
- Default: 3 retries (4 total attempts), 1s base delay, 10s max delay
- Context-aware: respects cancellation during retry sleep
- Structured logging on each retry attempt and on final success after
retry
All existing tests continue to pass. New tests cover:
- Backoff calculation (increase, cap)
- Retry success on first attempt (no unnecessary retries)
- Retry on transient failure (succeeds after N attempts)
- Exhausted retries (returns last error)
- Context cancellation during retry sleep
- Integration: SendNotification retries transient 500s
- Integration: all three endpoints retry independently
- Integration: permanent failure exhausts retries
closes #62
