Release 0.2.0

Added

• Custom charset support via shared JSON definition files (shared/charsets/)
• load_charset(name_or_path) — load a bundled preset (e.g. chinese) or custom JSON file
• resolve_charsets(charset_names, charset_files) — merge base + extra charsets with deduplication
• Bundled charset presets: base, chinese, japanese, korean
• CLI --charset flag (repeatable) to enable preset charsets
• CLI --charset-file flag (repeatable) to load custom charset JSON files
• APDEV_EXTRA_CHARS environment variable — comma-separated charset names or file paths, used as fallback when no CLI args provided
• check-chars: Block Elements (U+2580-U+259F) and Braille Patterns (U+2800-U+28FF) to default allowed ranges

Changed

• Character ranges and dangerous codepoints now loaded from charsets/base.json instead of hardcoded constants
• check_file() and check_paths() accept optional extra_ranges and dangerous kwargs for custom charset support
• is_allowed_char() now excludes dangerous codepoints (Trojan Source vectors) even though they fall within the General Punctuation allowed range

Fixed

• is_allowed_char() previously returned True for dangerous codepoints (U+200B, U+202E, etc.) because they fall within the General Punctuation range (U+2000-U+206F)
• load_charset() now catches all exceptions from importlib.resources (not just FileNotFoundError)
• check_file() now uses lazy-loaded cache instead of re-parsing base.json on every call