### What happened?

We recently turned on the "usage-based-routing-v2" routing strategy but had to revert to simple-shuffle. We are seeing failed calls and the stacktrace leads to the strategy code at times not finding a deployment to use when there are clearly some.

### Relevant log output

```shell Traceback (most recent call last):\n File \"/usr/local/lib/python3.12/site-packages/litellm/router_strategy/lowest_tpm_rpm_v2.py\", line 504, in async_get_available_deployments\n assert deployment is not None\n ^^^^^^^^^^^^^^^^^^^^^^\nAssertionError\n\nDuring handling of the above exception, another exception occurred:\n\nTraceback (most recent call last):\n File \"/usr/local/lib/python3.12/site-packages/litellm/router.py\", line 7278, in async_get_available_deployment\n await self.lowesttpm_logger_v2.async_get_available_deployments(\n File \"/usr/local/lib/python3.12/site-packages/litellm/router_strategy/lowest_tpm_rpm_v2.py\", line 554, in async_get_available_deployments\n raise litellm.RateLimitError(\nlitellm.exceptions.RateLimitError: litellm.RateLimitError: No deployments available for selected model. Passed model=gpt-5. Deployments={'cf177c86187fcf97eaae4d…