commit: UTF-8 recognition patch

[Tomas Frydrych]

I have committed Andrew Dunbar's patch (thanks Andrew), with 
some small changes. Notably, I removed the 
UT_ASSERT(SHOULD_NOT_HAPPEN) that Vlad had misgivings 
about, for I think the branch in which the assert was located can be 
executed legally in 8-bit non-utf8 text files.

Could someone who uses 8-bit, non-latin1 locale test that they can 
still open their non-utf8 text files?

Tomas