Add Requesty as a built-in provider

[Thibaultjaigu]

Adds Requesty as a built-in provider, mirroring the existing OpenRouter entry as closely as possible.

Requesty (https://requesty.ai) is an OpenAI-compatible LLM router — base URL https://router.requesty.ai/v1, GET /v1/models for dynamic model discovery, provider/model naming, and optional HTTP-Referer / X-Title attribution headers, same as OpenRouter. It fits alongside the OpenRouter and LiteLLM entries Halo already ships.

Changes:

• src/shared/constants/providers.ts: new requesty entry in BUILTIN_PROVIDERS (apiUrl https://router.requesty.ai/v1, modelsUrl .../v1/models, a small seed model list, icon: 'route'), placed next to OpenRouter in the global section.
• src/shared/types/ai-sources.ts: 'requesty' added to the BuiltinProviderId union.
• src/main/openai-compat-router/server/provider-adapters.ts: a requestyAdapter mirroring openRouterAdapter (matches on router.requesty.ai, injects the same optional HTTP-Referer / X-Title headers), registered in the adapters list.

Seed model ids used are verified live on Requesty (openai/gpt-4o-mini, openai/gpt-4o, anthropic/claude-sonnet-4-5, google/gemini-2.5-flash, deepseek/deepseek-chat); the full list is fetched dynamically from /v1/models.

I work at Requesty. This mirrors the existing OpenRouter provider as closely as possible. Happy to adjust or close it if it's not a fit.

[openkursar-flynn]

Thanks for the PR — this is a clean integration. The provider entry, type union, and adapter all mirror the OpenRouter pattern exactly the way we'd want, and we're happy to have Requesty as a built-in provider.

One small change requested before merge: please remove recommended: true from the provider entry. We reserve the recommended flag for providers with established community adoption, and as a policy we don't set it on vendor-submitted integrations — nothing personal, the same rule will apply to anyone.

Once that's updated we'll merge. Thanks again for the well-crafted contribution, and welcome aboard!