docs: document OpenAI Responses API and Anthropic Messages API routing

## Summary

KServe added HTTPRoute rules for the **OpenAI Responses API** (`/v1/responses`) in v0.18 and the **Anthropic Messages API** (`/v1/messages`) in v0.20 for LLMInferenceService. Neither endpoint is documented.

## What needs to be documented

- `/v1/responses` endpoint routing through InferencePool (OpenAI Responses API)
- `/v1/messages` endpoint routing through InferencePool (Anthropic Messages API)
- Both follow the same HTTPRoute pattern as `/v1/chat/completions` and `/v1/completions`
- Model-based routing rules are automatically included for both endpoints
- Update any existing endpoint/API reference tables to include these endpoints

## Related PRs

- https://github.com/kserve/kserve/pull/5291 (OpenAI Responses API, v0.18)
- https://github.com/kserve/kserve/pull/5648 (Anthropic Messages API, v0.20)

## Since

KServe v0.18 (`/v1/responses`), v0.20 (`/v1/messages`)

Search the documentation for the existing HTTPRoute examples and endpoint/API reference tables covering /v1/chat/completions and /v1/completions. Follow that structure to document /v1/responses and /v1/messages, including model-based routing, and confirm both endpoints appear in the relevant references.

Written by the indexing model from the issue text.