- Add a vocabulary to allow setting distractor pairs for common text terms matching phrase terms, to do early pruning of pairs of text tokens and phrase tokens.
- Switch to using `python-levelshtein` for faster Levenshtein computation and early stopping.
- Add an option to pad text and phrase tokens with boundary characters (``) to increase matches when one of the beginning or ending characters matches, or for very short words (shorter than ngram size).
**Full Changelog**: https://github.com/marijnkoolen/fuzzy-search/compare/v2.3.0...v2.4.0