Handle TimeoutError gracefully to prevent blocking HA startup

[sebr]

Summary

Fixes #370

• Wraps TimeoutError from aiohttp requests in RequestError at the API client level (client.py), converting it into the BHyveError hierarchy so all existing error handlers catch it automatically
• Adds defense-in-depth TimeoutError handling at the integration setup (__init__.py), coordinator (coordinator.py), and options flow (config_flow.py) layers
• Transient API timeouts during startup now raise ConfigEntryNotReady, which tells HA to retry with exponential backoff instead of blocking startup
• Transient API timeouts during polling raise UpdateFailed, which preserves the last known cached data and retries at the next coordinator interval — entities stay available with slightly stale data

Test plan

• Existing tests pass (133 passed, 5 skipped)
• Linting passes
• Verify HA starts normally when B-hyve API is temporarily unreachable
• Verify entities recover and update once the API becomes available again