fix(proxy): preserve reasoning_content for tool calls

[codeasier]

Summary / 概述

Preserve assistant thinking blocks as OpenAI-compatible reasoning_content when converting Anthropic messages with tool_use into Chat Completions format. Moonshot AI rejects assistant tool call history when thinking is enabled but reasoning_content is missing, so this also injects a small fallback for tool call messages without thinking content.

Related Issue / 关联 Issue

Fixes #2365

Screenshots / 截图

Before / 修改前

After / 修改后

Checklist / 检查清单

• pnpm typecheck passes / 通过 TypeScript 类型检查 (not run; Rust-only change)
• pnpm format:check passes / 通过代码格式检查 (not run; cargo fmt --check passed)
• cargo clippy passes (if Rust code changed) / 通过 Clippy 检查（如修改了 Rust 代码）
• Updated i18n files if user-facing text changed / 如修改了用户可见文本，已更新国际化文件 (no user-facing text changed)

Verification / 验证

• cargo fmt --check
• cargo clippy --lib -- -D warnings
• cargo test --lib transform

[farion1231]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8c288377e1

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[codeasier]

Addressed the review feedback in 160b587. The converter now only emits assistant messages when text/image content or tool_calls exist, so thinking-only blocks continue to be skipped instead of becoming empty assistant turns. Added test_anthropic_to_openai_skips_thinking_only_message. Verified with cargo fmt --check, cargo clippy --lib -- -D warnings, and cargo test --lib transform.