implements pattern for retrieving canonical identifiers in gpad output

This work is in reference to geneontology/pathways2GO#71

This allows entity ontologies aside from neo to be used to construct go-cams while maintaining GPAD outputs that adhere strictly to canonical terminologies such as UniProt for human genes.  It works by adding the annotation property  http://geneontology.org/lego/canonical_record to link new terms (e.g. reactome entities) to canonical terms (e.g. corresponding uniprots).  When these annotations are present, the GPAD SPARQL export process begins by converting the model to one with all of the external types replaced by canonical types.  The rest of the gpad export process is then unchanged.

minerva-cli/src/main/java/org/geneontology/minerva/cli/MinervaCommandRunner.java

@@ -419,7 +419,7 @@ else if (opts.nextEq("--ontology")) {
 		BlazegraphMolecularModelManager<Void> m3 = new BlazegraphMolecularModelManager<>(ontology, curieHandler, modelIdPrefix, inputDB, null);
 		for (IRI modelIRI : m3.getAvailableModelIds()) {
 			try {
-				String gpad = new GPADSPARQLExport(curieHandler, m3.getLegacyRelationShorthandIndex(), m3.getTboxShorthandIndex(), m3.getDoNotAnnotateSubset()).exportGPAD(m3.createInferredModel(modelIRI));
+				String gpad = new GPADSPARQLExport(curieHandler, m3.getLegacyRelationShorthandIndex(), m3.getTboxShorthandIndex(), m3.getDoNotAnnotateSubset()).exportGPAD(m3.createCanonicalInferredModel(modelIRI));
 				String fileName = StringUtils.replaceOnce(modelIRI.toString(), modelIdPrefix, "") + ".gpad";
 				Writer writer = new OutputStreamWriter(new FileOutputStream(Paths.get(gpadOutputFolder, fileName).toFile()), StandardCharsets.UTF_8);
 				writer.write(gpad);