numfmt: do not panic on a multibyte whitespace field separator

[leeewee]

Fixes #13482.

write_formatted_with_whitespace normalized the delimiter before the 2nd+ selected field by writing a single space and dropping the first byte of the leading whitespace run with &prefix[1..]. That assumed an ASCII separator, but WhitespaceSplitter splits on Unicode char::is_whitespace, so a multibyte separator (e.g. U+3000 IDEOGRAPHIC SPACE) made &prefix[1..] slice inside the character and abort:

$ numfmt --field 2 $'1　2'
thread 'main' panicked at src/uu/numfmt/src/format.rs:848:24:
byte index 1 is not a char boundary; it is inside '\u{3000}' (bytes 0..3) of `　`
$ echo $?
134

This drops the first character (by its UTF-8 length) instead of the first byte, so the slice always lands on a char boundary. Output becomes 1 2, exit 0.

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/tail/tail-n0f (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/cut/bounded-memory (passes in this run but fails in the 'main' branch)