fix(plugin-file-tree): fix 2-space indentation parsing for deeply nested tree structures

[SoonIter]

Summary

Related issues

AI Summary

---

What changes were made

1. Added detectIndentSize() function - Scans all lines to detect whether the tree uses 2-space or 4-space indentation mode
2. Updated calculateIndent() function - Now accepts an indentSize parameter to correctly calculate indent levels based on the detected mode
3. Added comprehensive test case - Tests a docs structure with 3-level nesting using 2-space indentation

Why they were made

The original tree parser had a bug when handling 2-space indented trees with 3 or more nesting levels. For example:

```tree
docs
├── _meta.json
└── guides
├── _meta.json
└── basic
├── introduction.mdx
├── install.mdx
└── plugin-development.md
```

The files under `basic` directory (4 spaces = 2+2) were incorrectly placed as siblings of `basic` instead of children, because the algorithm treated 4 spaces as a single 4-space indent unit rather than two 2-space units.

Implementation details

• The fix detects indentation mode by looking for lines with exactly 2 spaces before branch characters (`├──` or `└──`)
• Once 2-space mode is detected, all space-based indentation is calculated in 2-space units
• This ensures ` ├── file.mdx` (4 spaces) is correctly calculated as indent level 3 (2 units of 2-space + branch) instead of indent level 2

This PR was written using Vibe Kanban

---