docs: add LocalModelCache support for LLMInferenceService

## Summary

KServe v0.19 added LocalModelCache support for LLMInferenceService, but the existing LocalModelCache documentation (`docs/model-serving/generative-inference/modelcache/localmodel.md`) only covers usage with InferenceService.

## What needs to be documented

- How LocalModelCache integrates with LLMInferenceService
- Webhook defaulter behavior: automatic label/annotation setting when model URI matches a cache
- URI rewriting from `hf://model` to `pvc:///models//`
- Delete protection: cache deletion is blocked when active LLMInferenceServices reference it
- Label propagation to Deployment and LeaderWorkerSet pod templates
- Example YAML for LLMInferenceService using LocalModelCache and LocalModelNamespaceCache

## Related PR

- https://github.com/kserve/kserve/pull/5318

## Since

KServe v0.19

Update docs/model-serving/generative-inference/modelcache/localmodel.md, using the related PR #5318 as context for LocalModelCache support with LLMInferenceService. Document webhook defaults, URI rewriting, delete protection, label propagation, and YAML examples for LocalModelService and LocalModelNamespaceCache. Done means the existing documentation covers each listed behavior and example.

Written by the indexing model from the issue text.