Improve handling of unicode characters

[scampi]

Close #3574
Related to #6

[topecongiro]

LGTM.

A problem is that utils::unicode_str_width takes O(n) whereas String::len takes O(1). If we were to replace every s.len() with utils::unicode_str_width(s), then we may hit a performance issue.

One solution is to create our own string type which holds the width of the Grapheme Cluster and replace it with String, though this would require a significant amount of rewriting.