### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

## Summary

When using `Router` with multiple deployments under the same model group, a failed deployment may not be put into cooldown on later retry/fallback attempts.

From reading the code, it looks like Router cooldown depends on the sync failure logging callback:

- `Logging.failure_handler(...)` - `litellm.failure_callback` - `Router.deployment_callback_on_failure(...)` - `_set_cooldown_deployments(...)`

However, `Logging.failure_handler()` deduplicates `sync_failure` events using `has_logged_sync_failure` on `logging_obj.model_call_details`.

In Router retry/fallback flows, the same `litellm_logging_obj` appears to be reused across attempts. This means:

1. First provider/deployment failure runs `failure_handler()` normally. 2. `has_logged_sync_failure` is set to `True`. 3. Later retry/fallback provider failures call `failure_handler()` again, but the handler returns early due to deduplication. 4. `Router.deployment_callback_on_failure()` is not called for those later failed deployments. 5. Those failed deployments are not added to cool…