Lucene index returns incorrect results with has(SCORE_FIELD, lt(0.01d))

Nobody has claimed this yet.

Reported on [janusgraph-users](https://groups.google.com/d/msg/janusgraph-users/BW6aq14NSxg/ho85mhxhAQAJ) by Kirk T.

```
gremlin> GRAPH_DIR = "/tmp/tmp_main/graph";
==>/tmp/tmp_main/graph
gremlin> INDEX_DIR = "/tmp/tmp_main/index";
==>/tmp/tmp_main/index
gremlin> SCORE_FIELD = "myScore";
==>myScore
gremlin> SCORE_INDEX = "scoreIndex";
==>scoreIndex
gremlin> INDEX_BACKEND = "mySearch";
==>mySearch
gremlin> graph = JanusGraphFactory.build().
......1> set("storage.backend", "berkeleyje").
......2> set("storage.directory", GRAPH_DIR).
......3> set("index." + INDEX_BACKEND + ".backend", "lucene").
......4> set("index." + INDEX_BACKEND + ".directory", INDEX_DIR).
......5> open();
==>standardjanusgraph[berkeleyje:/tmp/tmp_main/graph]
gremlin> graph.tx().rollback();
==>null
gremlin> mgmt = graph.openManagement();
==>org.janusgraph.graphdb.database.management.ManagementSystem@5a8816cc
gremlin> scoreProp = mgmt.makePropertyKey(SCORE_FIELD).
......1> dataType(Double.class).
......2> cardinality(Cardinality.SINGLE).make();
==>myScore
gremlin> mgmt.buildIndex(SCORE_INDEX, Vertex.class).
......1> addKey(scoreProp).
......2> buildMixedIndex(INDEX_BACKEND);
==>scoreIndex
gremlin> mgmt.commit();
==>null
gremlin> graph.addVertex(SCORE_FIELD, 0d);
==>v[4192]
gremlin> graph.tx().commit();
==>null
gremlin> Thread.sleep(5*1000)
==>null
gremlin> g = graph.traversal();
==>graphtraversalsource[standardjanusgraph[berkeleyje:/tmp/tmp_main/graph], standard]
gremlin> g.V().has(SCORE_FIELD, P.gt(-1d)).toList(); // should return vertex
==>v[4192]
gremlin> g.V().has(SCORE_FIELD, P.lt(1d)).toList(); // should return vertex, but does not
gremlin> g.V().has(SCORE_FIELD, P.between(-1d, 1d)).toList(); // should return vertex, but does not
gremlin>
```

The final traversal strategy is:

```
Final Traversal [JanusGraphStep([],[myScore.lt(1.0)])]
```

The problem does not occur with Elasticsearch.

Start by running the supplied Gremlin reproduction with a Lucene mixed index and compare the lt and between traversals with the gt case. Trace the Lucene index path that handles numeric range predicates; done means the lt(1d) and between(-1d, 1d) traversals return the vertex, matching the Elasticsearch behavior.

Written by the indexing model from the issue text.