fix(filesystem): preserve UTF-8 across head and tail chunks

[Excelius-Wang]

Description

Fixes #4666.

Preserve multi-byte UTF-8 characters when headFile() and tailFile() read across 1024-byte chunk boundaries.

• Use StringDecoder for forward chunk decoding in headFile().
• Keep reverse-read chunks as bytes in tailFile() and decode only after enough line separators have been collected.
• Add regression tests with a CJK character split across the chunk boundary in both directions.

Server Details

• Server: filesystem
• Changes to: head_file and tail_file

Motivation and Context

Decoding every chunk independently replaces split UTF-8 sequences with U+FFFD. This corrupts CJK and other multi-byte text returned by the filesystem server.

How Has This Been Tested?

• npm test — 154 tests passed
• npm run build
• Added regression tests for headFile() and tailFile() with 界 split across a 1024-byte boundary

The change was tested at the filesystem operation boundary rather than through an LLM client.

Breaking Changes

None.

Types of changes

• Bug fix
• New feature
• Breaking change
• Documentation update

Checklist

• The code follows the repository style
• New and existing tests pass locally
• No README update is required because the public interface and configuration are unchanged
• No environment variables or configuration options were added

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.