Clarify why GIT_PYTHON_GIT_EXECUTABLE may be set on failure

This clarifies the comment that explains the significance of
setting the Git.GIT_PYTHON_GIT_EXECUTABLE attribute when running
Git failed but the attribute wasn't set before.

This happens in the code path in Git.refresh where has_git is False
and old_git is None, provided the refresh mode doesn't call for the
initial refresh to raise an exception on failure.

It was previously described in terms of a "first" and "second"
import, but as discussed in #1804, the rationale and effect were
not altogether clear.

Author: EliahKagan

git/cmd.py

@@ -475,9 +475,10 @@ def refresh(cls, path: Union[None, PathLike] = None) -> bool:
                     )
                     raise ImportError(err)
 
-                # We get here if this was the init refresh and the refresh mode was not
-                # error. Go ahead and set the GIT_PYTHON_GIT_EXECUTABLE such that we
-                # discern the difference between a first import and a second import.
+                # We get here if this was the initial refresh and the refresh mode was
+                # not error. Go ahead and set the GIT_PYTHON_GIT_EXECUTABLE such that we
+                # discern the the difference between the first refresh at import time
+                # and subsequent calls to refresh.
                 cls.GIT_PYTHON_GIT_EXECUTABLE = cls.git_exec_name
             else:
                 # After the first refresh (when GIT_PYTHON_GIT_EXECUTABLE is no longer