v0.0.3

Version 0.0.3

This is a milestone release with significant improvements in accuracy, usability, etc.

Improved

• Better accuracy thanks to updated trigrams (trained on new data) and script patterns
• Future proof script patterns that cover even more obscure scripts
• Chinese, Japanese, Korean (CJK) differentiation improved
• Bosnian, Croatian, Serbian differentiation improved
• Afrikaans, Dutch differentiation improved
• Improved detection for short texts (3-100 characters)
• Improved detection for casual/informal text (social media, chats, etc)