feat(manager): add OrcaRouter provider preset

[Marc-oss-hub]

Summary

Adds OrcaRouter as a named provider preset in the Manager's provider picker, mirroring the existing OpenRouter preset. OrcaRouter is an OpenAI-compatible model routing gateway that exposes models from OpenAI, Anthropic, Google, DeepSeek, Qwen and others behind a single endpoint and API key, with namespaced model IDs such as openai/gpt-5.5 and anthropic/claude-sonnet-4.6. It also runs gateway-level, zero-trust security for AI agents on the same endpoint — screening every prompt/response and governing every tool call on a default-deny basis, with no application code changes.

Picking the "OrcaRouter" preset fills in the provider name, Base URL (https://api.orcarouter.ai/v1), Chat Completions protocol, default model (openai/gpt-5.5) and a starter model list, so no manual endpoint typing is needed.

What this changes

• apps/codex-plus-manager/src/presets.ts: adds an orcarouter preset in the aggregator section, alongside OpenRouter and the other gateway presets. The preset selector auto-derives the provider patch (name / base URL / protocol / model / model list / relay mode) from the existing PRESETS entries, so no other wiring is required.

Why a separate preset rather than the OpenAI-compatible escape hatch

This mirrors how OpenRouter is already wired in this repo — a named aggregator preset that users can discover and pick directly, with its model list pre-filled. OrcaRouter requires namespaced model IDs (bare names are rejected by the gateway with 503 No available channel), so the preset sets openai/gpt-5.5 as the default and lists namespaced models in modelList.

How to use it

1. Get an API key at https://www.orcarouter.ai/console (keys start with sk-orca-).
2. In Codex++ Manager, open the provider picker → "从预设模板创建" → select OrcaRouter.
3. Paste the API key and connect. Models such as openai/gpt-5.5, anthropic/claude-sonnet-4.6, google/gemini-3.1-pro-preview and deepseek/deepseek-v4-flash are pre-listed; orcarouter/auto is an adaptive router that picks an upstream model per request.

Verification

• npm run check (tsc --noEmit) passes.
• npm test passes: 36 tests / 36 pass.
• npm run vite:build passes.
• Live API check against https://api.orcarouter.ai/v1/chat/completions with model openai/gpt-5.5 and an sk-orca- key returned HTTP 200 with the expected reply.

I'm an engineer on the OrcaRouter team.