bug: restart-ice must route to other server in case media-node stopped

According to bellow code, it is in TODO

https://github.com/8xFF/atm0s-media-server/blob/d632c3f90eafb246f15e17f28ffe6a019a9864f3/bin/src/server/gateway/local_rpc_handler.rs#L339-L360

The issue points to a TODO in gateway/local_rpc_handler.rs lines 339-360. Start by examining that file to understand the restart-ice logic and how media-node failures are handled. Look for related code that manages server routing and media-node health. Determine what 'route to other server' entails and how to detect a stopped media-node. Testing will involve simulating a media-node failure and verifying the restart-ice request is correctly rerouted.

Written by the indexing model from the issue text.