docs(api): add cached_tokens/reasoning_tokens to get_session() llm_token_usage example

[r266-tech]

#2305 added cached_tokens and reasoning_tokens to the llm_token_usage mapping on SessionMeta (default dict + from_dict in openviking/session/session.py), which is serialized via to_dict() and returned by get_session(). The get_session() Response Example in docs/{en,zh}/api/05-sessions.md still lists only prompt_tokens / completion_tokens / total_tokens, so the documented response shape is now incomplete.

This adds the two keys to the example (EN + ZH mirror) so the fields are discoverable — they let callers attribute prompt-cache savings (cached_tokens) and reasoning cost (reasoning_tokens) when reading get_session() output for LLM cost accounting.

Docs-only, scoped strictly to the get_session() llm_token_usage example block.

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

🎫 Ticket compliance analysis ❌ 2305 - Not compliant Non-compliant requirements: ov 增加处理的的reasoning、cache token统计 评测结果统计 所有token消耗和打印 增加api_key 控制不同的配置 增加飞书的卡片消息识别

⏱️ Estimated effort to review: 1 🔵⚪⚪⚪⚪

🏅 Score: 100

🧪 No relevant tests

🔒 No security concerns identified

✅ No TODO sections

🔀 No multiple PR themes

⚡ No major issues detected

[github-actions]

PR Code Suggestions ✨

No code suggestions found for the PR.