`longest_common_prefix` could fail with non-ascii codepoints

[Boscop]

longest_common_prefix finds the byte index of the longest common byte prefix:
https://docs.rs/commands/0.0.5/src/commands/util.rs.html#23-39

But slicing a string at a byte index that's not a char index (i.e. is in the middle of a utf-8 codepoint) is invalid:
https://stackoverflow.com/questions/51982999/slice-a-string-containing-unicode-chars
The function would panic if two of the given strings contain codepoints that start with the same byte but differ after that.
Using char_indices would fix this :)

[waywardmonkeys]

Would you be interested in submitting a PR?