Add BitRouter provider

[SPIKESPIGEL404]

What

Adds BitRouter as a provider. BitRouter is an OpenAI-compatible LLM gateway, so it uses the generic @ai-sdk/openai-compatible package — no new SDK dependency.

name = "BitRouter"
env = ["BITROUTER_API_KEY"]
npm = "@ai-sdk/openai-compatible"
api = "https://api.bitrouter.ai/v1"
doc = "https://docs.bitrouter.ai"

Models

50 models, mirrored from the live https://api.bitrouter.ai/v1/models catalog:

• 46 factored onto existing canonical base_models, overriding only [cost] and [limit] with BitRouter's live pricing/limits (same pattern as providers/kilo / providers/anyapi).
• 4 standalone where no canonical base exists yet (deepseek/deepseek-v3.2, openai/gpt-oss-120b, x-ai/grok-4.1-fast, x-ai/grok-4.20).

Reasoning

Each model declares [interleaved] field = "reasoning_content". Verified live against the gateway: BitRouter normalizes upstream reasoning traces to the reasoning_content field on the Chat Completions envelope (both streamed output and replayed back on input), which matches the openai-compatible reasoning path.

Validation

bun run validate passes.

[rekram1-node]

AUTOMATED REVIEW: Blocking merge: add the provider logo; add explicit provider-specific reasoning_options to every reasoning model; refresh prices against the live catalog; and factor Grok 4.20 through canonical base_model. The public mutable catalog should also have a sync implementation.