Instead of threading the state down to the larger loop, let's have the
loop where we detect the double star so each of them are easier to read.

In order to match the star-star, we disable the flag that's looking for
a single path element, but that leads to searching for the pattern in
the middle of elements in the input string.

Mark when we're handing a star-star so we jump over the elements in our
attempt to match the part of the pattern that comes after the star-star.

While here, tighten up the check so we don't allow invalid rules
through.

Add missing ')' to callbacks documentation

Super minor, but it was bugging me.

There was a missing closing paren in the docs.

tests: skip the unreadable file tests as root

When running as root, skip the unreadable file tests, because, well,
they're probably _not_ unreadable to root unless you've got some
crazy NSA clearance-level honoring operating system shit going on.

refs: provide a more general error message for dwim

Strict object creation in `refs::create`

When we turned strict object creation validation on by default, we
forgot to inform the refs::create tests of this.  They, in fact,
believed that strict object creation was off by default.  As a result,
their cleanup function went and turned strict object creation off for
the remaining tests.

This lets us run with strict object creation on.

If we cannot dwim the input, set the error message to be explicit about
that. Otherwise we leave the error for the last failed lookup, which
can be rather unexpected as it mentions a remote when the user thought
they were trying to look up a branch.

tests: fix core/stream test when built with openssl off

When passing -DUSE_OPENSSL:BOOL=OFF to cmake the testsuite will
fail with the following error:

core::stream::register_tls [/tmp/libgit2/tests/core/stream.c:40]
  Function call failed: (error)
  error -1 - <no message>

Fix test to assume failure for tls when built without openssl.
While at it also fix GIT_WIN32 cpp to check if it's defined
or not.

Remove Makefile.embed

This has not been a supported build mode for quite some time, and it
correspondingly hasn't worked to build the library for a long time. Get
rid of it, as the only build mode we support is though CMakek.

iterator/diff: allow trailing `/` on start/end paths to match submodules

ignore: don't use realpath to canonicalize path

Allow callers to specify a start path with a trailing slash to match
a submodule, instead of just a directory.  This is for some legacy
behavior that's sort of dumb, but there it is.

Test that submodules are found when the are included in a pathspec
but have a trailing slash.

If we're looking for a symlink, realpath will give us the resolved path,
which is not what we're after, but a canonicalized version of the path
the user asked for.

merge_driver: fix missing `goto done;`

The code initializing the merge driver registry accidentally
forgot a `goto done` in case of an error. Because of this the
next line, which registers the global shutdown callback for the
merge drivers, is only called when an error occured.

Fix this by adding the missing `goto done`. This fixes some
memory leaks when the global state is shut down.

WD iterator: properly identify submodules

Plug some leaks

Add a no-op size_t typedef for the doc parser

Remove duplicated calls to git_mwindow_close

xprepare memleaks

Clang's documentation parser, which we use in our documentation system
does not report any comments for functions which use size_t as a type.

The root cause is buried somewhere in libclang but we can work around it
by defining the type ourselves. This typedef makes sure that libclang
sees it and that we do not change its size.

The xdl_prepare_env() function may initialise an xdlclassifier_t
data structure via xdl_init_classifier(), which allocates memory
to several fields, for example 'rchash', 'rcrecs' and 'ncha'.
If this function later exits due to the failure of xdl_optimize_ctxs(),
then this xdlclassifier_t structure, and the memory allocated to it,
is not cleaned up.

In order to fix the memory leak, insert a call to xdl_free_classifier()
before returning.

This patch was originally written by Ramsay Jones (see commit
87f16258367a3b9a62663b11f898a4a6f3c19d31 in git.git).

Commit 307ab20b3 ("xdiff: PATIENCE/HISTOGRAM are not independent option
bits", 19-02-2012) introduced the XDF_DIFF_ALG() macro to access the
flag bits used to represent the diff algorithm requested. In addition,
code which had used explicit manipulation of the flag bits was changed
to use the macros.

However, one example of direct manipulation remains. Update this code to
use the XDF_DIFF_ALG() macro.

This patch was originally written by Ramsay Jones (see commit
5cd6978a9cfef58de061a9525f3678ade479564d in git.git).

config: don't write duplicate section

config: don't special-case multivars that don't exist yet