feat: add Requesty as an OpenAI-compatible LLM provider

[Thibaultjaigu]

Adds Requesty as a first-class, OpenAI-compatible LLM provider, mirroring the existing OpenRouter provider across the runtime, CLI, web UI, and docs.

Requesty (https://requesty.ai) is an OpenAI-compatible LLM gateway: one API across many models, using the same provider/model naming as OpenRouter, so it reuses the existing OpenAI-compatible request path.

Changes (mirrors OpenRouter at each wiring site)

Runtime (src/)

• types.ts — "requesty" added to the ProviderName union.
• provider.ts — PROVIDER_API_KEY_ENV (REQUESTY_API_KEY), catalog entry (base https://router.requesty.ai/v1, default model openai/gpt-4o-mini), display label, normalizeProvider, defaultBaseUrl, and both chat-completions routing branches.
• runtime/setup-api.ts — SUPPORTED_PROVIDERS / ENV_KEY_PROVIDERS / real-provider check.
• provider-capabilities.ts — context-window + modality cases (reusing the OpenRouter provider/model slug logic).
• model-routes.ts, embeddings.ts — route priority + OpenAI-embedding-compatible set.

CLI (src/cli/)

• commands/setup.ts — provider module + PROVIDERS / AUTO_CONFIGURABLE.
• commands/provider.ts, output.ts — name guard, default model, help text.

Web UI (web/src/)

• settings/_components/ProviderCard.tsx, components/ProviderPicker.tsx — selectable/removable provider sets + labels/descriptions.
• components/provider-logos.tsx, lib/providers.ts — icon + brand label.

Docs — docs/providers/requesty.md + index + README entries.

Testing

• bun run typecheck (root + web + mobile tsc --noEmit) — passes, exit 0.
• Verified live against the real endpoint: POST https://router.requesty.ai/v1/chat/completions with model: openai/gpt-4o-mini → HTTP 200 with a real completion.

Docs: https://docs.requesty.ai • Keys: https://app.requesty.ai/api-keys

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.