metric(qwen3): report prefix-cache query/hit

[Kokoro2336]

Description

Resolves #603

Report prefix-cache query/hit

Type of Change

• New feature (non-breaking change which adds functionality)

Checklist

• My code follows the style guidelines of this project (see docs/conventions/coding-style.md).

• I have performed a self-review of my own code.

• I have formatted my commits according to Commitizen conventions.

• I have run the local test suite and all tests pass (see CLAUDE.md).

[Kokoro2336]

I've made a simple test on the feature, sending the same prompt twice and filtering out queries/hits.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6bfd50749b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Expose a server feature for Qwen3 KV offload

When --kv-offload is passed to the server after this change, openinfer-qwen3 is normally compiled via the server's qwen3 feature, which only enables dep:openinfer-qwen3 and does not forward openinfer-qwen3/kv-offload. This fallback therefore rejects the run, and the suggested --features qwen3-kv-offload feature is not defined in openinfer-server/Cargo.toml (I checked the server feature list). Please add/forward a server feature or change the guidance; otherwise Qwen3 offload deployments cannot be built from the server path.

Useful? React with 👍 / 👎.

[xiaguan]

This PR makes openinfer-kv-offload optional without forwarding the feature from openinfer-server, so the default Qwen3 server rejects --kv-offload while recommending a nonexistent qwen3-kv-offload feature.