filesystem encoding master bugs

[Martin Pool]

After some discussion with mgz, I want to create/clarify the master
bugs for filesystem encoding issues, along the lines gz describes in
<https://bugs.launchpad.net/bzr/+bug/715547/comments/7>:

 * http://pad.lv/63324 - bzr can't handle files with filenames that
can't be be decoded to unicode
* http://pad.lv/794353 - bzr defaults to assuming the filesystem
encoding is ascii, which is usually wrong

It seems to me that if we solved bug 794353, it would fix most of the
filesystem unicode problems.  Not that many people really want to
version files that have names that are just arbitrary byte strings
with relation to characters.  Some do; if we implemented a
configuration option they might be better off just pretending it's all
8859-1 or something.

There are other specific bugs that even when the data is decodable and
we know the right encoding we still get it wrong, such as
http://pad.lv/715547.

Martin