Bug: Encoder output cannot be decoded by own decoder (indentation mismatch)

[acc0mplish]

Bug Description

The encoder creates TOON output that fails to parse with the decoder due to indentation mismatch.

Steps to Reproduce

# Encode JSON to TOON
echo '{"questions":[{"question":"백엔드","header":"백엔드","options":[{"label":"FastAPI","description":"Python"},{"label":"Django","
description":"Python"}],"multiSelect":false}]}' | toon

# Output (generated by encoder):
questions[1]:
  - question: 백엔드
    header: 백엔드
    options[2]{label,description}:
    FastAPI,Python
    Django,Python
    multiSelect: false

# Try to decode the same output
echo 'questions[1]:
  - question: 백엔드                                                                                                                
    header: 백엔드
    options[2]{label,description}:
    FastAPI,Python
    Django,Python
    multiSelect: false' | toon --decode

Error

Error: Parse error at line 5, column 12: Invalid indentation for tabular row: expected 6 spaces, found 4

Expected Behavior

The encoder output should be decodable by the decoder (round-trip compatibility).

Environment

- Rust: 1.92.0
- toon-format: 0.4.1
- OS: WSL2 (Linux)

Proposed Fix

The encoder and decoder should agree on indentation rules for tabular rows. Either:
1. Fix encoder to use 6 spaces for tabular rows
2. Fix decoder to accept 4 spaces for tabular rows

[shreyasbhat0]

fixed at #52