Completing Unicode (using \k regex) is broken because common prefix uses bytes instead of characters

[powerman]

Here is an example:

# Add some text to provide source words:
получаем битый символ
получив битый символ
# Then type "пол" and attempt to complete it result in adding broken unicode "учÐ":
получÐ

This happens because Cyrillic "а" in UTF8 is 0xD0 0xB0 and Cyrillic "и" in UTF8 is 0xD0 0xB8, so completion returns common part Latin Capital Letter Eth 0xD0.

[powerman]

Any chance it's a wrong repo for the issue and it should be in https://github.com/hrsh7th/nvim-cmp?