Verify jev_api_key (Option A) against the live jevai.org model-route API

[TheCoder30ec4]

Context

Routing with jev_api_key calls POST https://www.jevai.org/api/v1/decisions/model-route (choose_via_jev in src/model_router/jev.py). The request and response format was taken from Jev's docs and is covered by unit tests. But every live call so far returned HTTP 429 "Too many requests", so it has never been checked end to end. Option B (openrouter_api_key) is verified.

What to check with a working Jev key

• The request is accepted (task, candidates[{id, description}], priorities)
• data.decision is the chosen candidate id exactly as sent (ids contain / and .)
• 100+ candidates within the 32 KiB limit are accepted
• What a real rate-limit or error response looks like, so RouterError messages are useful
• Measured cost per routing call, to add next to the Option B figures in the README

Please post the (key-free!) request and response shapes you saw. Once verified, remove the caveat from the README's Limitations section.