Unicode String version 2.3.1

Bug Fixes

• Word breaking in a dictionary locale now applies the dictionary only to runs of text written in the script(s) that dictionary covers, with the standard Unicode rules governing everything else. Previously Unicode.String.split("Japanese", break: :word, locale: :ja) returned each letter separately.

• Word and line breaking no longer raise a File.Error when the ICU dictionaries have not been downloaded with mix unicode.string.download.dictionaries. Segmentation now falls back to the standard Unicode rules, and Unicode.String.break/2 and Unicode.String.splitter/2 return {:error, reason}.