docs: add llm-d multi-node inference guide

[EdisonSu768]

Summary

• explain the boundary between llm-d routing, LeaderWorkerSet workload management, and vLLM/NCCL model parallelism
• add a reusable two-node, four-GPU-per-node LLMInferenceService template using TP=4 and PP=2
• document prerequisites, deployment, real-request verification, and common scheduler, CRD, NCCL, backend, and KV-cache failures
• link the guide from the Inference Service and vLLM Expert Parallel pages

Validation

• yarn lint
• yarn build
• parsed the two-document YAML template
• validated both embedded vLLM launch scripts with bash -n

[cloudflare-workers-and-pages]

Deploying alauda-ai with    Cloudflare Pages

Latest commit:	ba86133
Status:	✅  Deploy successful!
Preview URL:	https://0d6bbce6.alauda-ai.pages.dev
Branch Preview URL:	https://docs-llm-d-multinode-inferen.alauda-ai.pages.dev

View logs