Fixing #215955 (UTF-8 caption corruption) revealed other UTF-8-related
issues.
1. Hardstatus/caption width accounting uses byte-length instead of
cell-width
This causes alignment/padding issues with non-ASCII captions.
2. OSC forwarding path re-encodes caption payload bytes in a way that
can
corrupt data that is already UTF-8.
3. After hardstatus redraw, cursor restore can be wrong in some
scenarios where
the caption includes UTF-8 chars, leading to prompt/cursor placement
anomalies until a redraw.
Minimal hardstatus to reproduce:
hardstatus alwayslastline "%=%t | %c "
A non-ASCII caption (e.g. Latin-extended or emojis) triggers incorrect
layout/cursor behaviour.
Tests:
printf '\e]0;%s\e\\' "abc ó ą ⭍ 💩"
printf '\e]0;%s\e\\' "Fußgängerübergänge"