fix: add aggressive fetching behaviour for first time for subnet_info

[shilingwang]

Summary

SubnetsInfoFetcher is a background task that periodically refreshes the NNS routing table and subnet-type snapshot. Previously it would wait for the full polling interval (default 5 min) between every attempt, even on startup when no snapshot had ever been fetched. This meant the gateway could be functionally unready for up to 5 minutes on a cold start if the first fetch failed.

This PR makes the first successful fetch aggressive: if SubnetsInfoFetcher.info is None (i.e. no snapshot has been populated yet), the run() method enters a tight retry loop with a 5-second back-off instead of returning and waiting for the normal interval. Once the first fetch succeeds the loop exits, the TaskManager interval takes over, and subsequent refreshes follow the configured --subnets-info-poll-interval as before.

Changes

src/routing/ic/subnets_info.rs
Added AGGRESSIVE_RETRY_INTERVAL constant (5 s).
Rewrote Run::run() to split into two paths:

• Healthy path (info is Some): single fetch, returns immediately — no behavioural change.

• Bootstrap path (info is None): retry loop with tokio::select! over the 5-second sleep and the shutdown CancellationToken, so it exits cleanly on service shutdown.