Libgit2 partially forwards OS error message texts.
On non-english Windows OSes these errors may contain non-ascii characters (i.e. umlauts).
To avoid that a UnicodeDecodeError is raised the error message is interpreted as UTF-8.
The solution should not be necessary on linux/osx as they return always ascii (as far as I know).
Thus this solution will not change the behaviour on linux/osx.
This represents what's going on much better than the remnants from the
older methods. What we do is pass a list of callbacks to libgit2 for it
to call, and they are valid for a single operation, not for the remote
itself.
This should also make it easier to re-use callbacks which have already
been set up.
This halves the amount of code we have to take into account for dealing
with the config.
There is a slight change in the API. Config.get_multivar() returns an
iterator instead of a list, which lets us reuse code from the general
iterator and is closer to libgit2's API.
This moves enough code into python with CFFI to pass the test_remotes
unit tests. There is no credentials support yet.
There is a small change in the return value of Remote.fetch() in that we
now return a TransferProgress object instead of extracting a few values
into a dictionary.
