Project name strips non-ASCII (CJK) characters from path, resulting in truncated/unrecognizable names

[XX888QM]

Bug Description

When indexing a project with a path containing non-ASCII characters (e.g. Chinese/CJK), the generated project name strips all non-ASCII characters and only retains the ASCII portions of the path.

Steps to Reproduce

1. Have a project at a path containing Chinese characters, e.g.:
   /Users/yunxin/Desktop/开发/后端/信租风控通后端
2. Run index_repository on this path
3. The generated project name becomes Users-yunxin-Desktop — all Chinese segments are silently dropped

Expected Behavior

The project name should either:

• Preserve Unicode characters as-is: Users-yunxin-Desktop-开发-后端-信租风控通后端
• Or percent-encode them: Users-yunxin-Desktop-%E5%BC%80%E5%8F%91-...
• Or allow users to provide a custom name parameter in index_repository to override the auto-generated name

Actual Behavior

All non-ASCII path segments are stripped. For the path above, the result is just Users-yunxin-Desktop, losing all meaningful project identification.

Impact

• Duplicate entries in the UI when the same project is indexed via both its real path and a symlink with an ASCII name
• Project names are meaningless for users with non-ASCII (Chinese, Japanese, Korean, etc.) directory structures
• No way to distinguish between multiple projects under the same ASCII-only parent path

Environment

• codebase-memory-mcp v0.8.1
• macOS (Apple Silicon)
• Project path with Chinese characters