Completions are offset when line contains non-ASCII characters

[TotalVerb]

I see that #1 has been fixed, but there still seems to be an annoying issue with non-ASCII characters. The autocompletion suggestions seem to be offset by some number of bytes when there are non-ASCII characters preceding it; see the screenshot attached (and note the part bolded). When I try to accept this suggestion, the result is momoved.

[NicoWeio]

This issue is still present for me. :/