str::char_at_reverse is broken

[ahnurmi]

char_at_reverse(s, i) returns the character at i - 1 instead of s.char_len() - 1 - i.

[huonw]

This is actually the "correct" behaviour (at least it is the behaviour listed in the documentation of char_range_at_reverse). Also, note that char_at and char_at_reverse actually deal with byte positions, not unicode char positions (this is not stated in the documentation for either of these :( ).

Pull request #5957 hopefully corrects both of these issues.

[z0w0]

Closing, has been clarified in a documentation update.