Add TeamoRouter: permanent free DeepSeek V4 Pro/Flash tiers

[blammectrappora]

Adds TeamoRouter to Inference providers (alphabetical, after SiliconFlow), following the existing entry format: description + footnote, Base URL, and a model table for the free-tier models.

Free tier: two permanent free model IDs granted at registration, no credit card:

Model	Context	Max Output	Rate Limit
deepseek-v4-flash-free	1M	384K	50 req/day
deepseek-v4-pro-free	1M	— (not published)	200 req/day

The -free IDs run the same model binaries as the paid tier — V4 Flash is the 284B MoE (MIT-licensed weights) with thinking mode on by default.

OpenAI SDK-compatible base URL https://api.teamorouter.cn/v1; the endpoint also accepts Anthropic Messages-format requests natively (works as a drop-in for Claude Code / Codex ANTHROPIC_BASE_URL/OPENAI_BASE_URL setups).

One caveat for your verification bar: the quota/context figures above are from the operator's published docs, and I have not yet run a live canary request against the free IDs (no key at hand while preparing this). If your process requires a live-request confirmation per row, happy to run one and report back — tell me the probe format you prefer.

Useful links:

• https://teamorouter.cn (landing page)
• https://teamorouter.cn/docs (API docs)
• https://api.teamorouter.cn/v1 (base URL)