feat(llm): add per-model reasoning effort

[allenxu09]

Summary

• store reasoning effort per model in backward-compatible model_settings
• send the protocol-specific request fields:
  • OpenAI Chat Completions: reasoning_effort
  • OpenAI Responses: reasoning.effort
  • Anthropic Messages: output_config.effort
• add reasoning-effort selection to the provider/model TUIs and VS Code configuration UI
• add CLI, environment, and GitHub Action overrides plus English/Chinese documentation

Why

Reasoning effort is model-specific, and each supported protocol uses a different request shape. Previously OCR had no first-class way to select it, so users had to rely on global extra_body overrides that did not follow model switches cleanly.

User impact

Users can select an effort while adding or switching models, set it with ocr config set reasoning_effort <value>, override it per run with --reasoning-effort or OCR_LLM_REASONING_EFFORT, and reset to the provider default without removing other model settings.

Validation

• go test ./internal/llm ./internal/agent
• targeted CLI/config/TUI tests in ./cmd/opencodereview
• go vet ./internal/llm ./internal/agent ./cmd/opencodereview
• VS Code Jest suite: 13 suites / 103 tests
• VS Code webpack compile and ESLint (one pre-existing warning in IdleView.tsx)
• npm run test:github-actions
• README and docs translation-sync checks

[CLAassistant]

All committers have signed the CLA.

[github-actions]

🔍 OpenCodeReview found 5 issue(s) in this PR.

• ✅ Successfully posted inline: 5 comment(s)

⚠️ 1 warning(s) occurred during review.

---

⚠️ Warnings:

• cmd/opencodereview/provider_tui.go (subtask_error): main_task did not complete before stopping

[allenxu09]

@lizhengfeng101 ping

[zhaozhenxiang]

这是一个刚性需求，需要实现。
我没有review pr的代码。

[allenxu09]

@MuoDoo @lizhengfeng101 ptal