Scale mediasoup across multiple Workers using pipeToRouter

Verified steps

1. Create one Worker per CPU core using mediasoup.createWorker() in a loop; keep a workers array and assign new rooms to the least-loaded Worker by tracking each Worker's router count.
2. When a producer on Router A needs to be consumed by a participant on Router B (different Worker or host), call routerA.pipeToRouter({ producerId, router: routerB }) which creates a Pipe transport pair and pipes RTP between the two Routers.
3. pipeToRouter returns { pipeConsumer, pipeProducer }; the pipeProducer on the destination Router can be consumed by participants on that Router the same way a regular producer is consumed.
4. For cross-host scaling, implement your own signaling to exchange the PipeTransport connect parameters between hosts, then call pipeTransport.connect() on each side with the remote IP, port, and SRTP parameters.
5. Monitor each Worker's resource use via worker.getResourceUsage() and redistribute rooms if a Worker approaches capacity (mediasoup documents a guideline of roughly 500 total consumers per Worker as a starting reference).
6. Handle Worker crashes with the worker 'died' event; restart the Worker and reconnect affected rooms — mediasoup does not auto-restart Workers.