Tweak bulk loader config handling

This is for compatibility with SparkGraphComputer.  In particular, I
seem to need the useGraphConfig method when executing on
SparkGraphComputer, even though it's not required on
GiraphGraphComputer.  I think {Spark,Giraph}GraphComputer may have
some kind of discrepancy in how they handle VP configs that bulk
loader is triggering (maybe bulk loader is doing something odd?)

titan-core/src/main/java/com/thinkaurelius/titan/graphdb/tinkerpop/computer/bulkloader/BulkLoaderVertexProgram.java

@@ -57,8 +57,15 @@ private BulkLoaderVertexProgram() {
 
     }
 
+    public BulkLoaderVertexProgram useGraphConfig(final Configuration c) {
+        this.configuration = c;
+        return this;
+    }
+
     public void storeState(final Configuration configuration) {
         VertexProgram.super.storeState(configuration);
+        if (this.configuration == null)
+            return;
         this.configuration.getKeys().forEachRemaining(key -> {
             configuration.setProperty(CFG_GRAPH_PREFIX + key, this.configuration.getProperty(key));
         });