The configured/base fetch refspecs need to be taken into account in
order to implement opportunistic remote-tracking branch updates. DWIM
them and store them in the struct, but don't do anything with them yet.

We can only DWIM when we've connected to the remote and have the list of
the remote's references. Adding or setting the refspecs should not
trigger an attempt to DWIM the refspecs as we typically cannot do it,
and even if we did, we would not use them for the current fetch.

This lets us test this bit as well as getting closer to what they were
trying to do.

With opportunistic ref updates, git has introduced the concept of having
base refspecs *and* refspecs that are active for a particular fetch.

Let's start by letting the user override the refspecs for download.

Stash ignored directories

Pool/Index data is not aligned

Correctly handle getaddrinfo return result

global: free the error message when exiting a thread

* Ensure alignment by copying the content into a structure on the stack

Remove unused buf variable from path/core test.

The getaddrinfo function indicates failure with a non-zero return code,
but this code is not necessarily negative. On platforms like Android
where the code is positive, a failed call causes libgit2 to segfault.

Factor 40 and 41 constants from source.

It looks like what Travis wants is a list of hashes, rather than a hash
with a list.

Fix `git_path_walk_up` to work with non-rooted paths

Replace void casts with GIT_UNUSED.

Provide a callback for certificate validation

travis: run valgrind only once

Instead of running valgrind on each job, half of which are in release
mode and don't have much usable information for valgrind, perform an
debug build as part of allowed_failures and run valgrind on that one,
which should speed up the feedback we get from the builds.

Skip it before we attempt to clone, as we would exit with -1 on systems
which do not have sshd running.

This brings us back in line with the other transports.