largefiles: fix explicit commit of normal/largefile switch

Commit of corresponding normal/largefiles pairs would only commit the standin.
That is usually fine, except if either the normal file or the standin is a
remove while the other is an add. In that case it would either give duplicate
colliding entries or lose the file.

Instead, commit both filenames if one of them is a remove.

hgext/largefiles/lfutil.py

@@ -562,8 +562,11 b' def updatestandinsbymatch(repo, match):'
     for f in match._files:
         fstandin = standin(f)
 
-        # ignore known largefiles and standins
-        if f in lfiles or fstandin in standins:
+        # For largefiles, only one of the normal and standin should be
+        # committed (except if one of them is a remove).
+        # Thus, skip plain largefile names but keep the standin.
+        if (f in lfiles or fstandin in standins) and \
+            repo.dirstate[f] != 'r' and repo.dirstate[fstandin] != 'r':
             continue
 
         actualfiles.append(f)

tests/test-largefiles.t

@@ -1850,7 +1850,7 b' the add and the remove is committed.'
   .hglf/normal3
   .hglf/sub2/large6
   .hglf/sub2/large7
-  normal3
+  large4-renamed
   sub/normal4
 
   $ cd ..