pygit2: implement status

[dtrifiro]

• bump pygit2 to 1.10.0
• force dulwich.status to return posix-style paths for untracked files

[skshetry]

pygit2 v1.10.0 is now released.

[dtrifiro]

Just some housekeeping

[dtrifiro]

While fixing this, I noticed something weird: since reset is not implemented for dulwich, in `estst:

scmrepo/tests/test_git.py

Lines 659 to 681 in f35b124

	@pytest.mark.skip_git_backend("pygit2")
	def test_add(tmp_dir: TmpDir, scm: Git, git: Git):
	tmp_dir.gen({"foo": "foo", "bar": "bar", "dir": {"baz": "baz"}})
	git.add(["foo", "dir"])
	staged, unstaged, untracked = scm.status()
	assert set(staged["add"]) == {"foo", "dir/baz"}
	assert len(unstaged) == 0
	assert len(untracked) == 1
	scm.commit("commit")
	tmp_dir.gen({"foo": "bar", "dir": {"baz": "bar"}})
	git.add([], update=True)
	staged, unstaged, _ = scm.status()
	assert set(staged["modify"]) == {"foo", "dir/baz"}
	assert len(unstaged) == 0
	assert len(untracked) == 1
	scm.reset()
	git.add(["dir"], update=True)
	staged, unstaged, _ = scm.status()
	assert set(staged["modify"]) == {"dir/baz"}
	assert len(unstaged) == 1
	assert len(untracked) == 1

The reset call switches backend from dulwich to pygit2, and since this PR adds pygit2.status, the scm.status call uses dulwich while the second call uses pygit2. I guess this is one of these issues that only come up when switching backends.

[skshetry]

@dtrifiro, in the future, please don't create a PR from your default branch. :)