This is just a bunch of small fixes that I noticed while looking
at the UTF8 and UTF16 path stuff.  It fixes a slowdown in looking
for an empty directory (not exiting loop asap), makes the dir name
in the git__DIR structure be a GIT_FLEX_ARRAY to save an allocation,
and fixes some slightly odd assumptions in the cl_getenv helper.

...and normalize the signatures of the two conversion functions.

Also fixed up call-sites to use the correct buffer sizes, especially
when converting to utf-8.

Fixes #1762

Allow Makefile.embed to be used when cross-compiling

This option is already present in the CMake config, but was missing from `Makefile.embed` and would cause all kinds of weird failures when compiling rugged on windows with the ruby devkit.

Fix possible double close

Fix -n bug; default to all ancestors

Signed-off-by: Sven Strickroth <email@cs-ware.de>

Close p->mwf.fd only if necessary

This fixes a regression introduced in revision 9d2f841a.

Signed-off-by: Sven Strickroth <email@cs-ware.de>

This allows libgit2 to be cross-compiled (e.g. when building native rugged binaries for windows from Linux or OS X).

```
CROSS_COMPILE=i686-w64-mingw32 make -f Makefile.embed
```

Doc fixes

The description of what the function does hasn't been true for quite a
while. Change it to reflect the way it currently works.

While here, remove an even older comment about missing features that
have been implemented.

clang's docparser highlighted these.

Invalid refs on disk cause revwalk globbing to fail

The new tests don't always want to use the same fixture data as
the old ones so this makes it configurable on a per-test basis.

Instead of using lots of strdup calls, this adds a memory pool to
the loose refs iteration code and uses it for keeping track of the
loose refs array.  Memory usage could probably be reduced even
further by eliminating the vector and just scanning by adding the
strlen of each ref, but that would be a more intrusive changes.

This also updates the error handling to be more thorough about
checking for failed allocations, etc.

The git_reference_next API silently skips invalid references when
scanning the loose refs.  The git_reference_next_name API should
skip the same ones even though it isn't creating the reference
object.

This adds a test with a an invalid loose reference and makes sure
that both APIs skip the same entries and generate the same results.

Clarify when to use github issues

git_buf_text_gather_stats doesn't work for multi-byte characters