DOC+TST: save: Update for "no commits" sub-repo fix in Git

Saving an untracked sub-repository without a commit checked out [0]
results in an ugly state: the sub-repository is _not_ added as a
submodule but its untracked files are added as blobs in the
_superdataset_.  This is not usually an issue for datasets because
these get initial commits on creation, so we documented this as a
known issue rather than introducing additional checks that would come
with a performance penalty.

The core problem is a combination of 'git submodule add' accepting a
sub-repository on an unborn branch and 'git ls-files' considering such
a sub-repository a directory rather than a repository.  Both issues
are fixed in Git 2.22.0 [1].  Update the documentation note and tests
accordingly.

[0] Usually this is just a repository without any commits, but it
    could be repository that has commits but is currently on an unborn
    branch.
[1] https://public-inbox.org/git/20190409230737.26809-1-kyle@kyleam.com/
    Commits e13811189b and b22827045e.

Re: #3139

datalad/core/local/save.py

@@ -81,10 +81,10 @@ class Save(Interface):
         % dataset save -d <path_to_dataset> --version-tag bestyet
 
     .. note::
-      For performance reasons, any Git repository without an initial commit
-      located inside a Dataset is ignored, and content underneath it will be
-      saved to the respective superdataset. DataLad datasets always have an
-      initial commit, hence are not affected by this behavior.
+      Before Git v2.22, any Git repository without an initial commit located
+      inside a Dataset is ignored, and content underneath it will be saved to
+      the respective superdataset. DataLad datasets always have an initial
+      commit, hence are not affected by this behavior.
     """
     # note above documents that out behavior is like that of `git add`, but
     # does not explicitly mention the connection to keep it simple.

datalad/core/local/tests/test_save.py

@@ -42,6 +42,7 @@
 from datalad.distribution.dataset import Dataset
 from datalad.support.annexrepo import AnnexRepo
 from datalad.support.exceptions import CommandError
+from datalad.support.external_versions import external_versions
 from datalad.api import (
     save,
     create,
@@ -662,18 +663,30 @@ def test_surprise_subds(path):
     # If subrepo is an adjusted branch, it would have a commit, making most of
     # this test irrelevant because it is about the unborn branch edge case.
     adjusted = somerepo.is_managed_branch()
+    # This edge case goes away with Git v2.22.0.
+    fixed_git = external_versions['cmd:git'] >= '2.22.0'
 
     # save non-recursive
-    ds.save(recursive=False)
+    res = ds.save(recursive=False, on_failure='ignore')
+    if not adjusted and fixed_git:
+        # We get an appropriate error about no commit being checked out.
+        assert_in_results(res, action='add_submodule', status='error')
+
     # the content of both subds and subrepo are not added to their
     # respective parent as no --recursive was given
     assert_repo_status(subds.path, untracked=['subfile'])
     assert_repo_status(somerepo.path, untracked=['subfile'])
 
-    if adjusted:
-        # adjusted branch: #datalad/3178 (that would have a commit)
-        modified = [subds.pathobj, somerepo.pathobj]
-        untracked = []
+    if adjusted or fixed_git:
+        if adjusted:
+            # adjusted branch: #datalad/3178 (that would have a commit)
+            modified = [subds.pathobj, somerepo.pathobj]
+            untracked = []
+        else:
+            # Newer Git versions refuse to add a sub-repository with no commits
+            # checked out.
+            modified = [subds.pathobj]
+            untracked = ['d1']
         assert_repo_status(ds.path, modified=modified, untracked=untracked)
         assert_not_in(ds.repo.pathobj / 'd1' / 'subrepo' / 'subfile',
                       ds.repo.get_content_info())