Performance improvements for GIC

- Made QueryProcessor filter columns that DNE
- Paralellized QueryProcessor::runQuery
- Switch `sout` logging to standard slf4j logging
  - Faster
  - No blocking threads due to synchronized methods

client-api/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/data/query/Query.java

@@ -4,9 +4,13 @@
 
 import edu.harvard.hms.dbmi.avillach.hpds.data.query.Filter.DoubleFilter;
 import edu.harvard.hms.dbmi.avillach.hpds.data.query.Filter.FloatFilter;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 public class Query {
 
+	private static final Logger log = LoggerFactory.getLogger(Query.class);
+
 	public Query() {
 
 	}
@@ -180,7 +184,7 @@ public String toString() {
 			break;
 		default:
 			//no logic here; all enum values should be present above
-			System.out.println("Formatting not supported for type " + expectedResultType);
+			log.warn("Formatting not supported for type {}", expectedResultType);
 		}
 
 		writePartFormat("Required Fields", requiredFields, builder, false);

processing/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/AbstractProcessor.java

@@ -627,7 +627,7 @@ public PhenoCube<?> load(String key) throws Exception {
 										inStream.close();
 										return ret;
 									}else {
-										System.out.println("ColumnMeta not found for : [" + key + "]");
+										log.warn("ColumnMeta not found for : [{}]", key);
 										return null;
 									}
 								}

processing/src/main/java/edu/harvard/hms/dbmi/avillach/hpds/processing/QueryProcessor.java

@@ -56,20 +56,22 @@ public String[] getHeaderRow(Query query) {
 	public void runQuery(Query query, AsyncResult result) {
 		TreeSet<Integer> idList = abstractProcessor.getPatientSubsetForQuery(query);
 		log.info("Processing " + idList.size() + " rows for result " + result.id);
-		for(List<Integer> list : Lists.partition(new ArrayList<>(idList), ID_BATCH_SIZE)){
-			result.stream.appendResultStore(buildResult(result, query, new TreeSet<Integer>(list)));			
-		};
+		Lists.partition(new ArrayList<>(idList), ID_BATCH_SIZE).parallelStream()
+			.map(list -> buildResult(result, query, new TreeSet<>(list)))
+			.sequential()
+			.forEach(result.stream::appendResultStore);
 	}
 
 
 	private ResultStore buildResult(AsyncResult result, Query query, TreeSet<Integer> ids) {
-		List<String> paths = query.getFields();
-		int columnCount = paths.size() + 1;
-
-		List<ColumnMeta> columns = paths.stream()
+		List<ColumnMeta> columns = query.getFields().stream()
 			.map(abstractProcessor.getDictionary()::get)
 			.filter(Objects::nonNull)
 			.collect(Collectors.toList());
+		List<String> paths = columns.stream()
+			.map(ColumnMeta::getName)
+			.collect(Collectors.toList());
+		int columnCount = paths.size() + 1;
 
 		ArrayList<Integer> columnIndex = abstractProcessor.useResidentCubesFirst(paths, columnCount);
 		ResultStore results = new ResultStore(result.id, columns, ids);