I added further improvement to previous patch.

If dfaexec() runs in non-UTF8 locales, length and wide character
representation are checked for all characters of a line in a input
string.  However, if matched early in the line, results for remaining
characters are wasted.

The new patch doesn't reuse both `mblen_buf' and `inputwcs' but stops
using them, and checks multibyte characters on demand.  It enables to
accomplish to speed-up for matched early and reduce required memories.

Norihiro