docs: add Chinese deep-dive on chunking algorithm

[AperturePlus]

Motivation

• Provide a concise Chinese reference that explains how the current chunking implementation works so Chinese-speaking contributors and users can understand indexing behavior and debugging semantics.

Description

• Add doc/CHUNKING_ALGORITHM.zh-CN.md which documents the end-to-end chunking flow (AST-first semantic chunking, token-limit handling, SmartChunkSplitter priorities and context prefixes, fixed-size fallback with overlaps, import extraction, output shapes and trade-offs).

Testing

• No automated tests were executed because this is a documentation-only change; the file was added and committed without modifying runtime code.

---

Codex Task