[PATCH] More regex microoptimization

[Paolo Bonzini]

> In the "real world" UTF-8 is now the
> predominent encoding for a locale

Well, I believe that ISO-8859 (plus people that leave LC_ALL=C or unset) is
still predominant, and in addition many UTF-8 regexes are optimized to
mb_cur_max == 1 (see optimize_utf8 in regcomp.c).  But if you think that the
range of optimizable regexes is overall a minority, that's another objection.
I will prepare another patch, you will choose the one to apply.

Paolo