Speed-up recursive by flushing index only once for all entries

The merge-recursive implementation in C inherited the invariant
that the on-file index file is written out and later read back
after any index operations and writing trees from the original
Python implementation.  But it was only because the original
implementation worked at the scripting level.

There is no need to write out the index file after handling
every path.

Signed-off-by: Alex Riesen <raa.lkml@gmail.com>
Signed-off-by: Junio C Hamano <junkio@cox.net>

merge-recursive.c

@@ -954,8 +954,6 @@ static int process_renames(struct path_list *a_renames,
 	path_list_clear(&a_by_dst, 0);
 	path_list_clear(&b_by_dst, 0);
 
-	if (cache_dirty)
-		flush_cache();
 	return clean_merge;
 }
 
@@ -1083,9 +1081,6 @@ static int process_entry(const char *path, struct stage_data *entry,
 	} else
 		die("Fatal merge failure, shouldn't happen.");
 
-	if (cache_dirty)
-		flush_cache();
-
 	return clean_merge;
 }
 
@@ -1133,6 +1128,8 @@ static int merge_trees(struct tree *head,
 			if (!process_entry(path, e, branch1, branch2))
 				clean = 0;
 		}
+		if (cache_dirty)
+			flush_cache();
 
 		path_list_clear(re_merge, 0);
 		path_list_clear(re_head, 0);