chat: avoid splitting surrogate pairs in read_file

[Martin Check (martincheck)]

Fixes #330999

This change keeps read_file line truncation from splitting a UTF-16 surrogate pair at the 2,000-code-unit boundary. When the boundary falls between a high and low surrogate, truncation moves back one code unit before appending the existing marker.

A regression test covers an astral character beginning at index 1,999.

Validation on 80b2a71e868059b3177b68b125310f2f690252b9 with Node.js 22.23.2 and @vscode/copilot-api 0.5.2:

• Focused readFile.spec.tsx: 35 passed
• npm run typecheck: passed
• npm run compile: passed
• Full Copilot unit suite: 406 files passed, 5 skipped; 8,916 tests passed, 129 skipped
• git diff --check: passed

[Copilot]

Pull request overview

Prevents read_file truncation from producing malformed Unicode when the limit intersects a UTF-16 surrogate pair.

Changes:

• Adjusts the truncation boundary to preserve surrogate pairs.
• Adds regression coverage for an astral character at the boundary.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
extensions/copilot/src/extension/tools/node/readFileTool.tsx	Safely adjusts the line truncation boundary.
extensions/copilot/src/extension/tools/node/test/readFile.spec.tsx	Tests surrogate-pair boundary handling.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[roblourens]

Thanks!