Code review comment for lp://qastaging/~jameinel/bzrtools/bug_431241

This potentially fixes bug #431241

Basically, 'string.whitespace' can contain non-ascii characters by a strange set of circumstances. And it seems that sometimes / always the readline completer works with unicode characters. This forces an upcast of the string, which fails because 0xa0 is not an ascii char.

0xa0 is apparently Non-breaking-space in iso-8859-1. I have not figured out the code path that injects that character, as running python interactively seems to not include it, but every time I run via 'bzr shell' it does show up.

It would be nice to get this into a 2.0.1 (or 2.0.0.1 ?) release, so it could be bundled with the 2.0.0 installer, but we could always roll it into the next one.