Add per-request rerank overrides to HTTP server

[ronakice]

Summary

• add optional request-level rerank overrides to the HTTP API
• key reranker reuse by effective config instead of a single startup-bound instance
• extend HTTP and schema tests for override and cache behavior

Testing

• python -m unittest test.test_cli_http test.test_cli_introspection

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 687c79683c

ℹ️ About Codex in GitHub

Codex has been enabled to automatically 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 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[branta42]

PR Comment Addressing Summary

Addressed 1 unresolved review thread.

File	Source	Verdict	Summary
src/rank_llm/api/runtime.py	chatgpt-codex-connector	FIXED	Added explicit override type validation so malformed override JSON now fails with a 400 validation error instead of reaching reranker backend selection.

CI failures fixed: none; PR checks were already green.

Local validation run:

• .venv/bin/pre-commit run --all-files
• .venv/bin/python -m unittest test.test_cli_http test.test_cli_introspection

Totals: FIXED=1, ACKNOWLEDGED=0, WON'T FIX=0, NOT APPLICABLE=0, NEEDS HUMAN INPUT=0.

Deferred items: none.

Open NEEDS HUMAN INPUT items: none.