Add retry with exponential backoff for webhook notifications

[roeimichael]

Problem

In webhook.py, if the HTTP request times out or gets a 5xx/429 response, it returns False and the notification is permanently lost. There is no retry mechanism, no queue for failed notifications, and no way to know a notification was dropped. Slack rate limiting (429) silently discards anomaly alerts.

What to do

1. Add retry logic to send_webhook() in src/context_analyzer_tool/notify/webhook.py:
   • 3 attempts with exponential backoff (1s, 2s, 4s delays)
   • For 429 responses, respect the Retry-After header
   • Log each retry attempt
2. (Optional stretch) Persist failed payloads to a queue table and retry them periodically

Files to look at

• src/context_analyzer_tool/notify/webhook.py, lines 181-230
• Consider using tenacity library or a simple manual retry loop with asyncio.sleep