Reasoning: instead encoding every character one by one as we
encounter them to use half their length as the correct offset,
we can simply encode the whole string at once as utf-16le and
work with that directly.
Dan suca
If Dan shared it with Traitor I'll not have to spend my time on this
Not a, sorry for not letting you sleep
k thx bye
Will this stay in history?
This special check treated some emojies as 3 characters long but
this shouldn't have actually been done, likely due to the old
regex matching more things as emoji than it should (which would
have count as 2 too, making up for 1+3 from the new is_emoji()).
